9873e18b75
Add loading of persisted blocks
2016-12-15 08:36:07 +01:00
d56b281006
Rename Iterator to Postings
2016-12-15 07:05:29 +01:00
c1acd3fe85
Advance buffered iterator correctly on seek
2016-12-14 21:43:27 +01:00
e561c91d53
Implement proper buffered iterator
...
This adds a proper duration based lookback buffer for series iterators
to allow advancing sequentially while remaining able to calculate time
aggregating functions such as `rate` backwards.
It uses an array ring buffer to minimize heap allocations for
potentially hundreds of thousands of series for a single query.
2016-12-14 21:14:44 +01:00
ca89080128
Misc fixes for initial Prometheus integration
2016-12-14 18:38:46 +01:00
725385ea05
Fix compareLabels, add test
2016-12-14 15:47:05 +01:00
fc992fafc2
Change querier interface, initial implementations
2016-12-14 15:39:23 +01:00
6eeb0ef01c
Add queriers and partial mocks
2016-12-13 15:26:58 +01:00
9b400b4c58
Add chunk based series iterator
2016-12-12 19:12:55 +01:00
b334c3ade8
Write chunk skiplist and add series reader
2016-12-12 15:39:55 +01:00
ae379f385b
Fix label index write and add read path
2016-12-12 11:38:43 +01:00
10943b6d88
Add initial index reader implementation
2016-12-12 08:12:19 +01:00
70a0224f19
Change chunk sample number to BigEndian
2016-12-12 08:11:53 +01:00
81b4d570ad
Add series file reader
2016-12-11 15:54:25 +01:00
5e02e28f9c
Add proper mmap calls
2016-12-11 15:49:36 +01:00
8425df035d
Fix hashmap serialization
2016-12-11 15:49:24 +01:00
14dbc59f2b
cleanup and switching removal of unsafe calls.
2016-12-10 18:09:57 +01:00
eb9af096f9
Write hashmap pointers, simplify section writer
2016-12-10 10:13:54 +01:00
3a528c3078
Write plain postings list index
2016-12-10 09:44:00 +01:00
4eba874b04
Factor out section writer
2016-12-09 22:36:31 +01:00
0b77a3dafc
Write series references into index
2016-12-09 22:27:43 +01:00
55b36ab413
Index persistence fixes, write label index hash table
2016-12-09 22:12:16 +01:00
8cbc95c316
Write label value indices
2016-12-09 21:40:38 +01:00
1e0edf367b
Write index with symbol table
2016-12-09 21:23:34 +01:00
40a451694f
Refactor persistence into interfaces
2016-12-09 20:45:46 +01:00
62f9dc311c
misc
2016-12-09 16:54:38 +01:00
74f8dfd95d
Persist blocks periodically
2016-12-09 13:41:38 +01:00
0cf8bb9e53
Move sub-indexes into single index structure
2016-12-09 10:41:51 +01:00
8aa99a3ebd
misc
2016-12-09 10:00:14 +01:00
2c34a15fe6
Add initial seriailization of block data
2016-12-08 17:43:10 +01:00
3ef7da33c8
Restructure files
2016-12-08 12:21:03 +01:00
63b887eb62
Add Makefile
2016-12-08 12:00:05 +01:00
b845f8d3a1
Reduce test data allocations
2016-12-08 11:59:54 +01:00
ce82bdb71a
Add write benchmark utility
2016-12-07 17:30:10 +01:00
52276c6966
Bucket samples before appending.
...
This pre-sorts samples into buckets before appending them to reduce
locking of shards.
2016-12-07 17:10:49 +01:00
c5945177fb
chunks: helper for bit range
2016-12-07 15:37:37 +01:00
9b459458d0
Docs and interface definitions
2016-12-05 21:26:19 +01:00
83574b1565
Add new interfaces and skeleton
2016-12-04 13:16:11 +01:00
ff29705571
Delete old benchmarking tool
2016-12-04 13:15:49 +01:00
6f93a699e6
wipe
2016-12-02 17:49:05 +01:00
0b6d621471
chunks: bring back lead/trail reuse, truncate incomplete sample
2016-11-30 23:01:01 +01:00
8c48dc2ca5
chunks: varint encoding in first full 64bit numbers
...
This saves about 7 bytes per chunk
2016-11-30 22:14:23 +01:00
5e76fd3126
chunks: drop shift, remove branch
...
This removes the need for the shift value and just calculates
everything on the fly through count.
Removes a branch condition.
2016-11-30 21:50:39 +01:00
c9ee572709
chunks: don't mutate in bstream iterator
...
This replaces mutation of underlying bytes in the iterated slice
with a shift counter, which is used when reading the head byte.
This is avoids having to copy the entire slice for every new iterator.
2016-11-30 19:39:22 +01:00
f392c01c12
chunks: add randomized test
2016-11-29 22:43:24 +01:00
fa181a34c1
chunks: cleanup anything but xor encoding
...
xor encoding is fast enough for our purposes and provides
very good compression.
We remove all other ones that partially don't support floats
for the sake of simplicity.
2016-11-29 22:02:58 +01:00
e67cf768dc
chunks: remove intermeidate copy from xor chunk
2016-11-20 16:24:46 +01:00
7874d28f32
chunks: implement xor encoding
2016-11-20 14:33:00 +01:00
342aa82505
Fix pages and index package
2016-11-15 15:53:48 +01:00
af36c89178
Make adapter work + comment it out
...
The adapter drags in a whole dependency chain of Prometheus internal
packages due to its storage's hideous structure.
With dependency flattening the adapter can actually be included into
Prometheus in a working way but it breaks package builds. Thus we
comment it out until we found a way to fix this contraption.
2016-11-15 11:46:17 +01:00
Fabian Reinartz