Commit Graph

13 Commits (af2a1cb10c89de496cd4309ac532624b34112d74)

Author SHA1 Message Date
Bryan Boreham 90e98e0235 tsdb: create isolation transaction slice on demand
When Prometheus restarts it creates every series read in from the WAL,
but many of those series will be finished, and never receive any more
samples. By defering allocation of the txRing slice to when it is first
needed, we save 32 bytes per stale series.

Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
2023-10-21 13:45:47 +00:00
Bryan Boreham 6fe8217ce4 tsdb: shrink txRing with smaller integers
4 billion active transactions ought to be enough for anyone.

Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
2023-10-21 12:44:34 +00:00
Matthieu MOREL c8e7f95a3c ci(lint): enable predeclared linter
Signed-off-by: Matthieu MOREL <matthieu.morel35@gmail.com>
2023-05-21 07:33:54 +00:00
Matthieu MOREL fb3eb21230 enable gocritic, unconvert and unused linters
Signed-off-by: Matthieu MOREL <matthieu.morel35@gmail.com>
2023-04-13 19:20:22 +00:00
Bryan Boreham ff00dee262
tsdb: turn off transaction isolation for head compaction (#11317)
* tsdb: add a basic test for read/write isolation

* tsdb: store the min time with isolationAppender
So that we can see when appending has moved past a certain point in time.

* tsdb: allow RangeHead to have isolation disabled
This will be used when for head compaction.

* tsdb: do head compaction with isolation disabled
This saves a lot of work tracking appends done while compaction is ongoing.

Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
2022-09-27 19:31:23 +05:30
Bryan Boreham 136f8b0ebb
tsdb: comment reason for isolation tracking reads (#11301)
I find it useful to know why a restriction exists, to check whether that
reason still applies, or in which other places it might apply.

This is based on the note here:
https://github.com/prometheus/prometheus/pull/9270#pullrequestreview-743820956
on the PR where the original comment was added.

Signed-off-by: Bryan Boreham <bjboreham@gmail.com>

Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
2022-09-13 11:11:03 +02:00
Darshan Chaudhary 9dcf8b2208
Add the ability to disable tsdb isolation (#9270)
* Disable isolation in isolation struct

Signed-off-by: darshanime <deathbullet@gmail.com>

* Run tsdb tests with isolation disabled

Signed-off-by: darshanime <deathbullet@gmail.com>

* Check for isolation disabled in isoState.Close()

Signed-off-by: darshanime <deathbullet@gmail.com>

* use t.Skip to skip isolation tests when disabled

Signed-off-by: darshanime <deathbullet@gmail.com>

* address review comments

Signed-off-by: darshanime <deathbullet@gmail.com>

* fix test for defaultIsolationState

Signed-off-by: darshanime <deathbullet@gmail.com>

* Change flag name. Set flag in DB. Do not init txRing. Close isoState.

Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com>

* Test disabled isolation in CircleCI test_go

Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com>

* Skip isolation related tests in db_test.go

Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com>

Co-authored-by: Ganesh Vernekar <ganeshvern@gmail.com>
2021-11-19 15:41:32 +05:30
Darshan Chaudhary c4f2e9eec5
Add present_over_time (#9097)
* Add present_over_time

Signed-off-by: darshanime <deathbullet@gmail.com>

* Add tests for present_over_time

Signed-off-by: darshanime <deathbullet@gmail.com>

* Address PR comments

Signed-off-by: darshanime <deathbullet@gmail.com>

* Add documentation for present_over_time

Signed-off-by: darshanime <deathbullet@gmail.com>

* Update documentation

Signed-off-by: darshanime <deathbullet@gmail.com>

* Update documentation comment

Signed-off-by: darshanime <deathbullet@gmail.com>
2021-07-29 12:38:11 +02:00
Ganesh Vernekar 59d02b5ef0
tsdb: Block Head GC till pending readers are done reading (#9081)
* tsdb: Block Head GC till pending readers are done reading

Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com>

* Fix review comments

Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com>

* Fix review comments 2

Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com>

* Fix the exclusiveness of maxt in WaitForPendingReadersInTimeRange

Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com>
2021-07-20 14:17:20 +05:30
Bryan Boreham dc8f505595
tsdb: coalesce lock/unlock operations for append (#9061)
Fetch the low watermark value under the same lock as we need for the
appender, rather than releasing then re-aquiring a lock on the same
Mutex.

Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
2021-07-07 18:58:20 +05:30
Peter Štibraný ff80690a6e
Optimise lowWatermark in Isolation (#7332)
* Track open appenders in doubly-linked list to make lowWatermark O(1).
* Use RW locks.
* Added BenchmarkIsolationWithState.

Signed-off-by: Peter Štibraný <peter.stibrany@grafana.com>
2020-06-03 20:09:05 +02:00
beorn7 d9af11e636 Do not attempt isolation for appendID == 0
Signed-off-by: beorn7 <beorn@grafana.com>
2020-03-01 02:48:35 +01:00
beorn7 7f30b0984d Implement isolation
This has been ported from https://github.com/prometheus/tsdb/pull/306.

Original implementation by @brian-brazil, explained in detail in the
2nd half of this talk:
https://promcon.io/2017-munich/talks/staleness-in-prometheus-2-0/

The implementation was then processed by @gouthamve into the PR linked
above. Relevant slide deck:
https://docs.google.com/presentation/d/1-ICg7PEmDHYcITykD2SR2xwg56Tzf4gr8zfz1OerY5Y/edit?usp=drivesdk

Signed-off-by: beorn7 <beorn@grafana.com>
Co-authored-by: Brian Brazil <brian.brazil@robustperception.io>
Co-authored-by: Goutham Veeramachaneni <gouthamve@gmail.com>
2020-02-28 14:18:39 +01:00