2022-09-20 17:05:50 +00:00
|
|
|
// Copyright 2021 The Prometheus Authors
|
|
|
|
|
// Licensed under the Apache License, Version 2.0 (the "License");
|
|
|
|
|
// you may not use this file except in compliance with the License.
|
|
|
|
|
// You may obtain a copy of the License at
|
|
|
|
|
//
|
|
|
|
|
// http://www.apache.org/licenses/LICENSE-2.0
|
|
|
|
|
//
|
|
|
|
|
// Unless required by applicable law or agreed to in writing, software
|
|
|
|
|
// distributed under the License is distributed on an "AS IS" BASIS,
|
|
|
|
|
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
|
|
|
// See the License for the specific language governing permissions and
|
|
|
|
|
// limitations under the License.
|
|
|
|
|
|
|
|
|
|
package tsdb
|
|
|
|
|
|
|
|
|
|
import (
|
2024-05-09 09:55:30 +00:00
|
|
|
"context"
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
"sync"
|
2022-09-20 17:05:50 +00:00
|
|
|
"testing"
|
|
|
|
|
|
|
|
|
|
"github.com/stretchr/testify/require"
|
|
|
|
|
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
"github.com/prometheus/prometheus/model/labels"
|
2022-09-20 17:05:50 +00:00
|
|
|
"github.com/prometheus/prometheus/tsdb/chunkenc"
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
"github.com/prometheus/prometheus/tsdb/chunks"
|
2022-09-20 17:05:50 +00:00
|
|
|
)
|
|
|
|
|
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
// TestMemSeries_chunk runs a series of tests on memSeries.chunk() calls.
|
|
|
|
|
// It will simulate various conditions to ensure all code paths in that function are covered.
|
|
|
|
|
func TestMemSeries_chunk(t *testing.T) {
|
|
|
|
|
const chunkRange int64 = 100
|
|
|
|
|
const chunkStep int64 = 5
|
|
|
|
|
|
|
|
|
|
appendSamples := func(t *testing.T, s *memSeries, start, end int64, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
for i := start; i < end; i += chunkStep {
|
|
|
|
|
ok, _ := s.append(i, float64(i), 0, chunkOpts{
|
|
|
|
|
chunkDiskMapper: cdm,
|
|
|
|
|
chunkRange: chunkRange,
|
|
|
|
|
samplesPerChunk: DefaultSamplesPerChunk,
|
|
|
|
|
})
|
|
|
|
|
require.True(t, ok, "sample append failed")
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
type setupFn func(*testing.T, *memSeries, *chunks.ChunkDiskMapper)
|
|
|
|
|
|
|
|
|
|
type callOutput uint8
|
|
|
|
|
const (
|
|
|
|
|
outOpenHeadChunk callOutput = iota // memSeries.chunk() call returned memSeries.headChunks with headChunk=true & isOpen=true
|
|
|
|
|
outClosedHeadChunk // memSeries.chunk() call returned memSeries.headChunks with headChunk=true & isOpen=false
|
|
|
|
|
outMmappedChunk // memSeries.chunk() call returned a chunk from memSeries.mmappedChunks with headChunk=false
|
|
|
|
|
outErr // memSeries.chunk() call returned an error
|
|
|
|
|
)
|
|
|
|
|
|
|
|
|
|
tests := []struct {
|
|
|
|
|
name string
|
|
|
|
|
setup setupFn // optional function called just before the test memSeries.chunk() call
|
|
|
|
|
inputID chunks.HeadChunkID // requested chunk id for memSeries.chunk() call
|
|
|
|
|
expected callOutput
|
|
|
|
|
}{
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=0 on empty memSeries",
|
|
|
|
|
inputID: 0,
|
|
|
|
|
expected: outErr,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=1 on empty memSeries",
|
|
|
|
|
inputID: 1,
|
|
|
|
|
expected: outErr,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "firstChunkID > ix",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange, cdm)
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Empty(t, s.mmappedChunks, "wrong number of mmappedChunks")
|
|
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Equal(t, int64(0), s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, chunkRange-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
s.firstChunkID = 5
|
|
|
|
|
},
|
|
|
|
|
inputID: 1,
|
|
|
|
|
expected: outErr,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=0 on memSeries with no mmapped chunks",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange, cdm)
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Empty(t, s.mmappedChunks, "wrong number of mmappedChunks")
|
|
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Equal(t, int64(0), s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, chunkRange-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
},
|
|
|
|
|
inputID: 0,
|
|
|
|
|
expected: outOpenHeadChunk,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=1 on memSeries with no mmapped chunks",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange, cdm)
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Empty(t, s.mmappedChunks, "wrong number of mmappedChunks")
|
|
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Equal(t, int64(0), s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, chunkRange-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
},
|
|
|
|
|
inputID: 1,
|
|
|
|
|
expected: outErr,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=10 on memSeries with no mmapped chunks",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange, cdm)
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Empty(t, s.mmappedChunks, "wrong number of mmappedChunks")
|
|
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Equal(t, int64(0), s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, chunkRange-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
},
|
|
|
|
|
inputID: 10,
|
|
|
|
|
expected: outErr,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=0 on memSeries with 3 mmapped chunks",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange*4, cdm)
|
|
|
|
|
s.mmapChunks(cdm)
|
|
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Equal(t, chunkRange*3, s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, (chunkRange*4)-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
},
|
|
|
|
|
inputID: 0,
|
|
|
|
|
expected: outMmappedChunk,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=1 on memSeries with 3 mmapped chunks",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange*4, cdm)
|
|
|
|
|
s.mmapChunks(cdm)
|
|
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Equal(t, chunkRange*3, s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, (chunkRange*4)-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
},
|
|
|
|
|
inputID: 1,
|
|
|
|
|
expected: outMmappedChunk,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=3 on memSeries with 3 mmapped chunks",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange*4, cdm)
|
|
|
|
|
s.mmapChunks(cdm)
|
|
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Equal(t, chunkRange*3, s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, (chunkRange*4)-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
},
|
|
|
|
|
inputID: 3,
|
|
|
|
|
expected: outOpenHeadChunk,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=0 on memSeries with 3 mmapped chunks and no headChunk",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange*4, cdm)
|
|
|
|
|
s.mmapChunks(cdm)
|
|
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Equal(t, chunkRange*3, s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, (chunkRange*4)-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
s.headChunks = nil
|
|
|
|
|
},
|
|
|
|
|
inputID: 0,
|
|
|
|
|
expected: outMmappedChunk,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=2 on memSeries with 3 mmapped chunks and no headChunk",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange*4, cdm)
|
|
|
|
|
s.mmapChunks(cdm)
|
|
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Equal(t, chunkRange*3, s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, (chunkRange*4)-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
s.headChunks = nil
|
|
|
|
|
},
|
|
|
|
|
inputID: 2,
|
|
|
|
|
expected: outMmappedChunk,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=3 on memSeries with 3 mmapped chunks and no headChunk",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange*4, cdm)
|
|
|
|
|
s.mmapChunks(cdm)
|
|
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Equal(t, chunkRange*3, s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, (chunkRange*4)-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
s.headChunks = nil
|
|
|
|
|
},
|
|
|
|
|
inputID: 3,
|
|
|
|
|
expected: outErr,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=1 on memSeries with 3 mmapped chunks and closed ChunkDiskMapper",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange*4, cdm)
|
|
|
|
|
s.mmapChunks(cdm)
|
|
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Equal(t, chunkRange*3, s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, (chunkRange*4)-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
cdm.Close()
|
|
|
|
|
},
|
|
|
|
|
inputID: 1,
|
|
|
|
|
expected: outErr,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=3 on memSeries with 3 mmapped chunks and closed ChunkDiskMapper",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange*4, cdm)
|
|
|
|
|
s.mmapChunks(cdm)
|
|
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Equal(t, chunkRange*3, s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, (chunkRange*4)-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
cdm.Close()
|
|
|
|
|
},
|
|
|
|
|
inputID: 3,
|
|
|
|
|
expected: outOpenHeadChunk,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=0 on memSeries with 3 head chunks and no mmapped chunks",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange*3, cdm)
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Empty(t, s.mmappedChunks, "wrong number of mmappedChunks")
|
|
|
|
|
require.Equal(t, 3, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Equal(t, int64(0), s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, (chunkRange*3)-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
},
|
|
|
|
|
inputID: 0,
|
|
|
|
|
expected: outClosedHeadChunk,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=1 on memSeries with 3 head chunks and no mmapped chunks",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange*3, cdm)
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Empty(t, s.mmappedChunks, "wrong number of mmappedChunks")
|
|
|
|
|
require.Equal(t, 3, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Equal(t, int64(0), s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, (chunkRange*3)-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
},
|
|
|
|
|
inputID: 1,
|
|
|
|
|
expected: outClosedHeadChunk,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=10 on memSeries with 3 head chunks and no mmapped chunks",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange*3, cdm)
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Empty(t, s.mmappedChunks, "wrong number of mmappedChunks")
|
|
|
|
|
require.Equal(t, 3, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Equal(t, int64(0), s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, (chunkRange*3)-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
},
|
|
|
|
|
inputID: 10,
|
|
|
|
|
expected: outErr,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=0 on memSeries with 3 head chunks and 3 mmapped chunks",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange*4, cdm)
|
|
|
|
|
s.mmapChunks(cdm)
|
|
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
|
|
|
|
|
appendSamples(t, s, chunkRange*4, chunkRange*6, cdm)
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 3, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
|
|
|
|
require.Equal(t, chunkRange*3, s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, (chunkRange*6)-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
},
|
|
|
|
|
inputID: 0,
|
|
|
|
|
expected: outMmappedChunk,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=2 on memSeries with 3 head chunks and 3 mmapped chunks",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange*4, cdm)
|
|
|
|
|
s.mmapChunks(cdm)
|
|
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
|
|
|
|
|
appendSamples(t, s, chunkRange*4, chunkRange*6, cdm)
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 3, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
|
|
|
|
require.Equal(t, chunkRange*3, s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, (chunkRange*6)-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
},
|
|
|
|
|
inputID: 2,
|
|
|
|
|
expected: outMmappedChunk,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=3 on memSeries with 3 head chunks and 3 mmapped chunks",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange*4, cdm)
|
|
|
|
|
s.mmapChunks(cdm)
|
|
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
|
|
|
|
|
appendSamples(t, s, chunkRange*4, chunkRange*6, cdm)
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 3, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
|
|
|
|
require.Equal(t, chunkRange*3, s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, (chunkRange*6)-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
},
|
|
|
|
|
inputID: 3,
|
|
|
|
|
expected: outClosedHeadChunk,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=5 on memSeries with 3 head chunks and 3 mmapped chunks",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange*4, cdm)
|
|
|
|
|
s.mmapChunks(cdm)
|
|
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
|
|
|
|
|
appendSamples(t, s, chunkRange*4, chunkRange*6, cdm)
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 3, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
|
|
|
|
require.Equal(t, chunkRange*3, s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, (chunkRange*6)-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
},
|
|
|
|
|
inputID: 5,
|
|
|
|
|
expected: outOpenHeadChunk,
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=6 on memSeries with 3 head chunks and 3 mmapped chunks",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange*4, cdm)
|
|
|
|
|
s.mmapChunks(cdm)
|
|
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
|
|
|
|
|
appendSamples(t, s, chunkRange*4, chunkRange*6, cdm)
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 3, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
|
|
|
|
require.Equal(t, chunkRange*3, s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, (chunkRange*6)-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
},
|
|
|
|
|
inputID: 6,
|
|
|
|
|
expected: outErr,
|
|
|
|
|
},
|
|
|
|
|
|
|
|
|
|
{
|
|
|
|
|
name: "call ix=10 on memSeries with 3 head chunks and 3 mmapped chunks",
|
|
|
|
|
setup: func(t *testing.T, s *memSeries, cdm *chunks.ChunkDiskMapper) {
|
|
|
|
|
appendSamples(t, s, 0, chunkRange*4, cdm)
|
|
|
|
|
s.mmapChunks(cdm)
|
|
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 1, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
|
|
|
|
|
appendSamples(t, s, chunkRange*4, chunkRange*6, cdm)
|
2023-12-07 11:35:01 +00:00
|
|
|
require.Equal(t, 3, s.headChunks.len(), "wrong number of headChunks")
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
require.Len(t, s.mmappedChunks, 3, "wrong number of mmappedChunks")
|
|
|
|
|
require.Equal(t, chunkRange*3, s.headChunks.oldest().minTime, "wrong minTime on last headChunks element")
|
|
|
|
|
require.Equal(t, (chunkRange*6)-chunkStep, s.headChunks.maxTime, "wrong maxTime on first headChunks element")
|
|
|
|
|
},
|
|
|
|
|
inputID: 10,
|
|
|
|
|
expected: outErr,
|
|
|
|
|
},
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
memChunkPool := &sync.Pool{
|
|
|
|
|
New: func() interface{} {
|
|
|
|
|
return &memChunk{}
|
|
|
|
|
},
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
for _, tc := range tests {
|
|
|
|
|
t.Run(tc.name, func(t *testing.T) {
|
|
|
|
|
dir := t.TempDir()
|
|
|
|
|
chunkDiskMapper, err := chunks.NewChunkDiskMapper(nil, dir, chunkenc.NewPool(), chunks.DefaultWriteBufferSize, chunks.DefaultWriteQueueSize)
|
|
|
|
|
require.NoError(t, err)
|
|
|
|
|
defer func() {
|
|
|
|
|
require.NoError(t, chunkDiskMapper.Close())
|
|
|
|
|
}()
|
|
|
|
|
|
2024-01-29 11:57:27 +00:00
|
|
|
series := newMemSeries(labels.EmptyLabels(), 1, 0, true)
|
Use a linked list for memSeries.headChunk (#11818)
Currently memSeries holds a single head chunk in-memory and a slice of mmapped chunks.
When append() is called on memSeries it might decide that a new headChunk is needed to use for given append() call.
If that happens it will first mmap existing head chunk and only after that happens it will create a new empty headChunk and continue appending
our sample to it.
Since appending samples uses write lock on memSeries no other read or write can happen until any append is completed.
When we have an append() that must create a new head chunk the whole memSeries is blocked until mmapping of existing head chunk finishes.
Mmapping itself uses a lock as it needs to be serialised, which means that the more chunks to mmap we have the longer each chunk might wait
for it to be mmapped.
If there's enough chunks that require mmapping some memSeries will be locked for long enough that it will start affecting
queries and scrapes.
Queries might timeout, since by default they have a 2 minute timeout set.
Scrapes will be blocked inside append() call, which means there will be a gap between samples. This will first affect range queries
or calls using rate() and such, since the time range requested in the query might have too few samples to calculate anything.
To avoid this we need to remove mmapping from append path, since mmapping is blocking.
But this means that when we cut a new head chunk we need to keep the old one around, so we can mmap it later.
This change makes memSeries.headChunk a linked list, memSeries.headChunk still points to the 'open' head chunk that receives new samples,
while older, yet to be mmapped, chunks are linked to it.
Mmapping is done on a schedule by iterating all memSeries one by one. Thanks to this we control when mmapping is done, since we trigger
it manually, which reduces the risk that it will have to compete for mmap locks with other chunks.
Signed-off-by: Łukasz Mierzwa <l.mierzwa@gmail.com>
2023-07-31 09:10:24 +00:00
|
|
|
|
|
|
|
|
if tc.setup != nil {
|
|
|
|
|
tc.setup(t, series, chunkDiskMapper)
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
chk, headChunk, isOpen, err := series.chunk(tc.inputID, chunkDiskMapper, memChunkPool)
|
|
|
|
|
switch tc.expected {
|
|
|
|
|
case outOpenHeadChunk:
|
|
|
|
|
require.NoError(t, err, "unexpected error")
|
|
|
|
|
require.True(t, headChunk, "expected a chunk with headChunk=true but got headChunk=%v", headChunk)
|
|
|
|
|
require.True(t, isOpen, "expected a chunk with isOpen=true but got isOpen=%v", isOpen)
|
|
|
|
|
case outClosedHeadChunk:
|
|
|
|
|
require.NoError(t, err, "unexpected error")
|
|
|
|
|
require.True(t, headChunk, "expected a chunk with headChunk=true but got headChunk=%v", headChunk)
|
|
|
|
|
require.False(t, isOpen, "expected a chunk with isOpen=false but got isOpen=%v", isOpen)
|
|
|
|
|
case outMmappedChunk:
|
|
|
|
|
require.NoError(t, err, "unexpected error")
|
|
|
|
|
require.False(t, headChunk, "expected a chunk with headChunk=false but got gc=%v", headChunk)
|
|
|
|
|
case outErr:
|
|
|
|
|
require.Nil(t, chk, "got a non-nil chunk reference returned with an error")
|
|
|
|
|
require.Error(t, err)
|
|
|
|
|
}
|
|
|
|
|
})
|
|
|
|
|
}
|
|
|
|
|
}
|
2024-05-09 09:55:30 +00:00
|
|
|
|
|
|
|
|
func TestHeadIndexReader_PostingsForLabelMatching(t *testing.T) {
|
|
|
|
|
testPostingsForLabelMatching(t, 0, func(t *testing.T, series []labels.Labels) IndexReader {
|
|
|
|
|
opts := DefaultHeadOptions()
|
|
|
|
|
opts.ChunkRange = 1000
|
|
|
|
|
opts.ChunkDirRoot = t.TempDir()
|
|
|
|
|
h, err := NewHead(nil, nil, nil, nil, opts, nil)
|
|
|
|
|
require.NoError(t, err)
|
|
|
|
|
t.Cleanup(func() {
|
|
|
|
|
require.NoError(t, h.Close())
|
|
|
|
|
})
|
|
|
|
|
app := h.Appender(context.Background())
|
|
|
|
|
for _, s := range series {
|
|
|
|
|
app.Append(0, s, 0, 0)
|
|
|
|
|
}
|
|
|
|
|
require.NoError(t, app.Commit())
|
|
|
|
|
|
|
|
|
|
ir, err := h.Index()
|
|
|
|
|
require.NoError(t, err)
|
|
|
|
|
return ir
|
|
|
|
|
})
|
|
|
|
|
}
|
