Commit Graph

28 Commits

Author SHA1 Message Date
Goutham Veeramachaneni 10d395259b Avoid creation of 0 sized segments. (#527)
If the corrupt segment is full, then we set donePages on open,
c59ed492b2/wal/wal.go (L235-L243)

Then when we try to repair, we set the segment to be a new segment but
we don't update the donePages: c59ed492b2/wal/wal.go (L334)

We we try to log to this, because donePages is full, we will never log
anything to this segment and create a new one: c59ed492b2/wal/wal.go (L486)

This does not cause issues because we simply concatenate the segments on
read, there by transparently skipping this `0b` segment.
2019-02-25 12:10:27 +02:00
Tom Wilkie 77d5a7d47a
LiveReader can get into an infinite loop on corrupt WALs. (#524)
Make WAL live tailer return EOF when the there is a half-written record at the end of the file.

Previously, this would cause an infinite loop as we ignored EOFs when filling the buffer.  We now differentiate between EOFs that read >0 bytes, and EOFs that didn't.

Add some more unit tests for tailing a corrupt WAL, and unify interfaces Reader and LiveReader for the purposes of testing.

Signed-off-by: Tom Wilkie <tom.wilkie@gmail.com>
2019-02-19 14:33:57 +00:00
Tom Wilkie bc3b0bd429
Test to corrupt segments mid-WAL, repair and check we can read the correct number of records. (#528)
Test to corrupt segments mid-WAL, repair and check we can read the correct number of records.

Make segmentBufReader pad short segments with zeros, and only advance curr segment index after fully reading segment.
2019-02-18 19:05:07 +00:00
Callum Styan 89ee5aaed4 clarify which segments are deleted when we find a corrupted segment (#522)
Signed-off-by: Callum Styan <callumstyan@gmail.com>
2019-02-14 05:44:19 +02:00
Callum Styan 3929359302
add live reader for WAL (#481)
* add live reader for WAL

Signed-off-by: Callum Styan <callumstyan@gmail.com>
2019-01-16 10:09:08 -08:00
Krasi Georgiev 8d991bdc1e
Delete temp checkpoint folder on error. (#415) 2019-01-07 11:43:33 +03:00
glutamatt 22e3aeb107 Add WALSegmentSize as an option of tsdb creation (#450)
Expose `WALSegmentSize` option to allow overriding the `DefaultOptions.WALSegmentSize`.
2018-12-18 21:56:51 +03:00
Krasi Georgiev 2962202ed3
fix windows tests (#469)
Signed-off-by: Krasi Georgiev <kgeorgie@redhat.com>
2018-12-13 16:29:29 +03:00
Krasi Georgiev 48efdf8b81
refactor NewSegmentsRangeReader to take multi WAL ranges (#449)
* refactor NewSegmentsRangeReader to take multi WAL ranges

In case of an error when checkpointing the WAL the error doesn't show
the exact WAL index that is corrupter. this is because it uses
MultiReader to read multiply WAL files.
This refactoring allows the NewSegmentsRangeReader to take more than a
single WAL range and it reads all of the ranges by iterating each one.

this changes the logs from
create checkpoint: read segments: corruption after 4841144384 bytes:...
to
create checkpoint: read segments: corruption in segment
data/wal/00017351 at 123142208: ...

Signed-off-by: Krasi Georgiev <kgeorgie@redhat.com>
2018-11-30 16:46:16 +02:00
Krasi Georgiev 0493efb7c5
repair wal when the record cannot be decoded (#453)
* repair wal when the record cannot be decoded

Currently repair is run only when the error happens in the reader.

A corruption can occur after the record is read and when it is decoded.
This change wraps the error at decoding as a CorruptionErr as this error
is expected to trigger a repair.

Signed-off-by: Krasi Georgiev <kgeorgie@redhat.com>
2018-11-30 13:37:04 +02:00
Krasi Georgiev 24520727a4
return an error when the last wal segment record is torn. (#451)
* return an error when the  last wal segment record is torn.

this ensures that a repair will be run when the last record in a segment
is torn.

Signed-off-by: Krasi Georgiev <kgeorgie@redhat.com>
2018-11-28 15:15:11 +02:00
Krasi Georgiev 3385571ddf
buffer-panic when reading a record after recPageTerm (#429)
Signed-off-by: Krasi Georgiev <kgeorgie@redhat.com>
2018-11-14 18:43:33 +02:00
Krasi Georgiev a9470dd8d5
few more comments to explain the WAL workflow (#430)
More comments for the WAL package.

Signed-off-by: Krasi Georgiev <kgeorgie@redhat.com>
2018-11-08 10:27:16 +02:00
Krasi Georgiev d7492b9350
more descriptive var names and some more logging. (#405)
* more descriptive checkpoint var names and some more logging.

Signed-off-by: Krasi Georgiev <kgeorgie@redhat.com>
2018-10-11 18:23:52 +03:00
Ganesh Vernekar 61b000ee0e
Fix review comments
Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2018-09-28 15:00:51 +05:30
Ganesh Vernekar 632dfb349e
Add new metrics.
1. 'prometheus_tsdb_wal_truncate_fail' for failed WAL truncation.
2. 'prometheus_tsdb_checkpoint_delete_fail' for failed old checkpoint delete.

Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
2018-09-25 18:50:57 +05:30
Goutham Veeramachaneni 9c8ca47399
Fix filehandling for windows (#392)
* Fix filehandling for windows

Signed-off-by: Goutham Veeramachaneni <gouthamve@gmail.com>

* Fix more windows filehandling issues

Windows: Close files before deleting Checkpoints.

Signed-off-by: Goutham Veeramachaneni <gouthamve@gmail.com>

Windows: Close writers in case of errors so they can be deleted

Signed-off-by: Goutham Veeramachaneni <gouthamve@gmail.com>

Windows: Close block so that it can be deleted.

Signed-off-by: Goutham Veeramachaneni <gouthamve@gmail.com>

Windows: Close file to delete it

Signed-off-by: Goutham Veeramachaneni <gouthamve@gmail.com>

Windows: Close dir so that it can be deleted.

Signed-off-by: Goutham Veeramachaneni <gouthamve@gmail.com>

Windows: close files so that they can be deleted.

Signed-off-by: Goutham Veeramachaneni <gouthamve@gmail.com>

* Review feedback

Signed-off-by: Goutham Veeramachaneni <gouthamve@gmail.com>
2018-09-21 11:01:22 +05:30
Goutham Veeramachaneni c7d0d10da4
Make sure WAL Repair can handle wrapped errors
Signed-off-by: Goutham Veeramachaneni <gouthamve@gmail.com>
2018-09-19 12:12:25 +05:30
beorn7 3bc6c670fa Revert "Remove `prometheus_` prefix from metrics"
This reverts commit 98fe30438c.

After some discussion, it was concluded that we want the full
`prometheus_tsdb_...` prefix hardcoded in the library.

Signed-off-by: beorn7 <beorn@soundcloud.com>
2018-09-18 19:19:19 +02:00
beorn7 98fe30438c Remove `prometheus_` prefix from metrics
This can now be added by users of the library as needed with the new
https://godoc.org/github.com/prometheus/client_golang/prometheus#WrapRegistererWithPrefix

Signed-off-by: beorn7 <beorn@soundcloud.com>
2018-09-17 14:54:28 +02:00
Fabian Reinartz 22cae653d8 Fixes for 32bit archs
Signed-off-by: Fabian Reinartz <freinartz@google.com>
2018-08-07 06:52:16 -04:00
Fabian Reinartz f8ec0074e7 Add Replace function
Signed-off-by: Fabian Reinartz <freinartz@google.com>
2018-08-02 17:51:49 -04:00
Fabian Reinartz b81e0fbf2a Address comments
Signed-off-by: Fabian Reinartz <freinartz@google.com>
2018-07-20 02:26:12 -04:00
Fabian Reinartz 3e76f0163e Address comments
Signed-off-by: Fabian Reinartz <freinartz@google.com>
2018-07-19 07:25:30 -04:00
Fabian Reinartz 3f538817f8 move WAL lock
Signed-off-by: Fabian Reinartz <freinartz@google.com>
2018-07-19 07:25:30 -04:00
Fabian Reinartz d951140ab8 wal: avoid heap allocation in WAL reader
The buffers we allocated were escaping to the heap, resulting in large
memory usage spikes during startup and checkpointing in Prometheus.
This attaches the buffer to the reader object to prevent this.

Signed-off-by: Fabian Reinartz <freinartz@google.com>
2018-07-19 07:25:30 -04:00
Fabian Reinartz 449a2d0db7 wal: add segment type and repair procedure
Allow to repair the WAL based on the error returned by a reader
during a full scan over all records.

Signed-off-by: Fabian Reinartz <freinartz@google.com>
2018-07-19 07:24:40 -04:00
Fabian Reinartz 8e1f97fad4 wal: add write ahead log package
This adds a new WAL that's agnostic to the actual record contents.
It's much simpler and should be more resilient than the existing one.

Signed-off-by: Fabian Reinartz <freinartz@google.com>
2018-07-19 07:24:40 -04:00