Add cleanup of the lockfile when the db is cleanly closed
The metric describes the status of the lockfile on startup
0: Already existed
1: Did not exist
-1: Disabled
Therefore, if the min value over time of this metric is 0, that means that executions have exited uncleanly
We can then use that metric to have a much lower threshold on the crashlooping alert:
If the metric exists and it has been zero, two restarts is enough to trigger the alarm
If it does not exist (old prom version for example), the current five restarts threshold remains
Signed-off-by: Julien Duchesne <julien.duchesne@grafana.com>
* Change metric name + set unset value to -1
Signed-off-by: Julien Duchesne <julien.duchesne@grafana.com>
* Only check the last value of the clean start alert
Signed-off-by: Julien Duchesne <julien.duchesne@grafana.com>
* Fix test + nit
Signed-off-by: Julien Duchesne <julien.duchesne@grafana.com>
* Added walreplay API endpoint
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Added starting page to react-ui
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Documented the new endpoint
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Fixed typos
Signed-off-by: Levi Harrison <git@leviharrison.dev>
Co-authored-by: Julius Volz <julius.volz@gmail.com>
* Removed logo
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Changed isResponding to isUnexpected
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Changed width of progress bar
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Changed width of progress bar
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Added DB stats object
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Updated starting page to work with new fields
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Passing nil
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Passing nil (pt. 2)
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Passing nil (pt. 3)
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Passing nil (and also implementing a method this time) (pt. 4)
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Passing nil (and also implementing a method this time) (pt. 5)
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Changed const to let
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Passing nil (pt. 6)
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Remove SetStats method
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Added comma
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Changed api
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Changed to triple equals
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Fixed data response types
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Don't return pointer
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Changed version
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Fixed interface issue
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Fixed pointer
Signed-off-by: Levi Harrison <git@leviharrison.dev>
* Fixed copying lock value error
Signed-off-by: Levi Harrison <git@leviharrison.dev>
Co-authored-by: Julius Volz <julius.volz@gmail.com>
* Added test to reproduce panic on TSDB head chunks truncated while querying
Signed-off-by: Marco Pracucci <marco@pracucci.com>
* Added test for Querier too
Signed-off-by: Marco Pracucci <marco@pracucci.com>
* Stop the bleed on mmap-ed head chunks panic
Signed-off-by: Marco Pracucci <marco@pracucci.com>
* Lower memory pressure in tests to ensure it doesn't OOM
Signed-off-by: Marco Pracucci <marco@pracucci.com>
* Skip TestQuerier_ShouldNotPanicIfHeadChunkIsTruncatedWhileReadingQueriedChunks
Signed-off-by: Marco Pracucci <marco@pracucci.com>
* Experiment to not trigger runtime.GC() continuously
Signed-off-by: Marco Pracucci <marco@pracucci.com>
* Try to fix test in CI
Signed-off-by: Marco Pracucci <marco@pracucci.com>
* Do not call runtime.GC() at all
Signed-off-by: Marco Pracucci <marco@pracucci.com>
* I have no idea why it's failing in CI, skipping tests
Signed-off-by: Marco Pracucci <marco@pracucci.com>
The main branch tests are not passing due to the fact that #8489 was not
rebased on top of #8007.
Signed-off-by: Julien Pivotto <roidelapluie@inuits.eu>
This moves the label lookup into TSDB, whilst still keeping the cached-ref optimisation for repeated Appends.
This makes the API easier to consume and implement. In particular this change is motivated by the scrape-time-aggregation work, which I don't think is possible to implement without it as it needs access to label values.
Signed-off-by: Tom Wilkie <tom.wilkie@gmail.com>
* CleanupTombstones refactored, now reloading blocks after every compaction.
The goal is to remove deletable blocks after every compaction and, thus, decrease disk space used when cleaning tombstones.
Signed-off-by: arthursens <arthursens2005@gmail.com>
* Protect DB against parallel reloads
Signed-off-by: ArthurSens <arthursens2005@gmail.com>
* Fix typos
Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
Co-authored-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
* fix: remove pre-2.21 tmp blocks on start
Signed-off-by: Nguyen Le Vu Long <vulongvn98@gmail.com>
* fix: commenting
Signed-off-by: Nguyen Le Vu Long <vulongvn98@gmail.com>
* Testify: move to require
Moving testify to require to fail tests early in case of errors.
Signed-off-by: Julien Pivotto <roidelapluie@inuits.eu>
* More moves
Signed-off-by: Julien Pivotto <roidelapluie@inuits.eu>
* Don't use returned DB to close resources on TSDB startup error
Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
* Add unit test and fix another panic
Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
* Fix review comment
Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
* Refactor test assertions
This pull request gets rid of assert.True where possible to use
fine-grained assertions.
Signed-off-by: Julien Pivotto <roidelapluie@inuits.eu>
* tsdb: Bug fix for further continued after crash deletions; added more tests.
Additionally: Added log line for block removal.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* Addressed comment.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
## Changes:
* Rename dir when deleting
* Ignoring blocks with broken meta.json on start (we do that on reload)
* Compactor writes <ulid>.tmp-for-creation blocks instead of just .tmp
* Delete tmp-for-creation and tmp-for-deletion blocks during DB open.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* tsdb: Added ChunkQueryable implementations to db; unified compactor, querier and fanout block iterating.
Chained to https://github.com/prometheus/prometheus/pull/7059
* NewMerge(Chunk)Querier now takies multiple primaries allowing tsdb DB code to use it.
* Added single SeriesEntry / ChunkEntry for all series implementations.
* Unified all vertical, and non vertical for compact and querying to single
merge series / chunk sets by reusing VerticalSeriesMergeFunc for overlapping algorithm (same logic as before)
* Added block (Base/Chunk/)Querier for block querying. We then use populateAndTomb(Base/Chunk/) to iterate over chunks or samples.
* Refactored endpoint tests and querier tests to include subtests.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* Addressed comments from Brian and Beorn.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* Fixed snapshot test and added chunk iterator support for DBReadOnly.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* Fixed race when iterating over Ats first.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* Fixed tests.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* Fixed populate block tests.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* Fixed endpoints test.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* Fixed test.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* Added test & fixed case of head open chunk.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* Fixed DBReadOnly tests and bug producing 1 sample chunks.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* Added cases for partial block overlap for multiple full chunks.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* Added extra tests for chunk meta after compaction.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* Fixed small vertical merge bug and added more tests for that.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* tsdb/chunks: Replace sync/atomic with uber-go/atomic
Signed-off-by: Javier Palomo <javier.palomo.almena@gmail.com>
* tsdb/heaad: Replace sync/atomic with uber-go/atomic
Signed-off-by: Javier Palomo <javier.palomo.almena@gmail.com>
* vendor: Make go.uber.org/atomic a direct dependency
There is no modifications to go.sum and vendor/ because
it was already vendored.
Signed-off-by: Javier Palomo <javier.palomo.almena@gmail.com>
* tsdb: Remove comments referring to the sync/atomic alignment bug
Related: https://golang.org/pkg/sync/atomic/#pkg-note-BUG
Signed-off-by: Javier Palomo <javier.palomo.almena@gmail.com>
* no panic the head memseries has chunks in it
Signed-off-by: Krasi Georgiev <8903888+krasi-georgiev@users.noreply.github.com>
* fix a panic when querying after a wal corruption.
Signed-off-by: Krasi Georgiev <8903888+krasi-georgiev@users.noreply.github.com>
* review nits
Signed-off-by: Krasi Georgiev <8903888+krasi-georgiev@users.noreply.github.com>
* Add test for reading the data after a wal corruption.
Signed-off-by: Krasi Georgiev <8903888+krasi-georgiev@users.noreply.github.com>
Update tsdb/db_test.go
Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
Update tsdb/db_test.go
Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
Signed-off-by: Krasi Georgiev <8903888+krasi-georgiev@users.noreply.github.com>
* spellings
Signed-off-by: Krasi Georgiev <8903888+krasi-georgiev@users.noreply.github.com>
Co-authored-by: Ganesh Vernekar <15064823+codesome@users.noreply.github.com>
* Add errors and Warnings to SeriesSet
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Change Querier interface and refactor accordingly
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Refactor promql/engine to propagate warnings at eval stage
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Address review issues
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Make sure all the series from all Selects are pre-advanced
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Address review issues
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Separate merge series sets
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Clean
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Refactor merge querier failure handling
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Refactored and simplified fanout with improvements from incoming chunk iterator PRs.
* Secondary logic is hidden, instead of weird failed series set logic we had.
* Fanout is well commented
* Fanout closing record all errors
* MergeQuerier improved API (clearer)
* deferredGenericMergeSeriesSet is not needed as we return no samples anyway for failed series sets (next = false).
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* Fix formatting
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Fix CI issues
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Added final tests for error handling.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* Addressed Brian's comments.
* Moved hints in populate to be allocated only when needed.
* Used sync.Once in secondary Querier to achieve all-or-nothing partial response logic.
* Select after first Next is done will panic.
NOTE: in lazySeriesSet in theory we could just panic, I think however we can
totally just return error, it will panic in expand anyway.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* Utilize errWithWarnings
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Fix recently introduced expansion issue
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Add tests for secondary querier error handling
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Implement lazy merge
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Add name to test cases
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Reorganize
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Address review comments
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Address review comments
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Remove redundant warnings
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
* Fix rebase mistake
Signed-off-by: Kemal Akkoyun <kakkoyun@gmail.com>
Co-authored-by: Bartlomiej Plotka <bwplotka@gmail.com>
When appending to the head and a chunk is full it is flushed to the disk and m-mapped (memory mapped) to free up memory
Prom startup now happens in these stages
- Iterate the m-maped chunks from disk and keep a map of series reference to its slice of mmapped chunks.
- Iterate the WAL as usual. Whenever we create a new series, look for it's mmapped chunks in the map created before and add it to that series.
If a head chunk is corrupted the currpted one and all chunks after that are deleted and the data after the corruption is recovered from the existing WAL which means that a corruption in m-mapped files results in NO data loss.
[Mmaped chunks format](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/head_chunks.md) - main difference is that the chunk for mmaping now also includes series reference because there is no index for mapping series to chunks.
[The block chunks](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/chunks.md) are accessed from the index which includes the offsets for the chunks in the chunks file - example - chunks of series ID have offsets 200, 500 etc in the chunk files.
In case of mmaped chunks, the offsets are stored in memory and accessed from that. During WAL replay, these offsets are restored by iterating all m-mapped chunks as stated above by matching the series id present in the chunk header and offset of that chunk in that file.
**Prombench results**
_WAL Replay_
1h Wal reply time
30% less wal reply time - 4m31 vs 3m36
2h Wal reply time
20% less wal reply time - 8m16 vs 7m
_Memory During WAL Replay_
High Churn:
10-15% less RAM - 32gb vs 28gb
20% less RAM after compaction 34gb vs 27gb
No Churn:
20-30% less RAM - 23gb vs 18gb
40% less RAM after compaction 32.5gb vs 20gb
Screenshots are in [this comment](https://github.com/prometheus/prometheus/pull/6679#issuecomment-621678932)
Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>
This is technically BREAKING CHANGE, but it was like this from the beginning: I just notice that we rely in
Prometheus on remote read being sorted. This is because we use selected data from remote reads in MergeSeriesSet
which rely on sorting.
I found during work on https://github.com/prometheus/prometheus/pull/5882 that
we do so many repetitions because of this, for not good reason. I think
I found a good balance between convenience and readability with just one method.
Smaller the interface = better.
Also I don't know what TestSelectSorted was testing, but now it's testing sorting.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
This fixes#6992, which was introduced by #6777. There was an
intermediate component which translated TSDB errors into storage errors,
but that component was deleted and this bug went unnoticed, until we
were watching at the Prombench results. Without this, scrape will fail
instead of dropping samples or using "Add" when the series have been
garbage collected.
Signed-off-by: Julien Pivotto <roidelapluie@inuits.eu>
This is technically BREAKING CHANGE, but it was like this from the beginning: I just notice that we rely in
Prometheus on remote read being sorted. This is because we use selected data from remote reads in MergeSeriesSet
which rely on sorting.
I found during work on https://github.com/prometheus/prometheus/pull/5882 that
we do so many repetitions because of this, for not good reason. I think
I found a good balance between convenience and readability with just one method.
Smaller the interface = better.
Also I don't know what TestSelectSorted was testing, but now it's testing sorting.
Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>
* tsdb: don't allow ingesting empty labelsets
When we ingest an empty labelset in the head, further blocks can not be
compacted, with the error:
```
level=error ts=2020-02-27T21:26:58.379Z caller=db.go:659 component=tsdb
msg="compaction failed" err="persist head block: write compaction:
add series: out-of-order series added with label set \"{}\" / prev:
\"{}\""
```
We should therefore reject those invalid empty labelsets upfront.
This can be reproduced with the following:
```
cat << END > prometheus.yml
scrape_configs:
- job_name: 'prometheus'
scrape_interval: 1s
basic_auth:
username: test
password: test
metric_relabel_configs:
- regex: ".*"
action: labeldrop
static_configs:
- targets:
- 127.0.1.1:9090
END
./prometheus --storage.tsdb.min-block-duration=1m
```
And wait a few minutes.
Signed-off-by: Julien Pivotto <roidelapluie@inuits.eu>
Ganesh Vernekar