[PERF] TSDB: Query head and ooo-head together
The current implementation of out-of-order querying runs two queriers which each iterate all series in the head, then runs a merge operation on the output.
This PR adds HeadAndOOOQuerier which iterates just once over series, then where necessary merges chunks from in-order and out-of-order lists.
In order to distinguish in-order from out-of-order chunk references I set bit 23 (i.e. 1<<23) on ooo references; this reduces the maximum number of chunks from 16 million to 8 million.
Note one side-effect of this change is that results may come in a different order - the merge operation done previously required a sort of all series. This only changes where Prometheus does not guarantee the order.
Fixes#11628
In `mmapCurrentOOOHeadChunk`, check if the number is at the maximum and
drop the data with an error log. This is not expected to happen as the
maximum is over 8 million; that's 8 years of 1 sample every second.
Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
* Re-enable check in `createHeadWithOOOSamples` which wasn't really broken.
* Move code making `Block` into a `Queryable` into test file.
* Make `getSeriesChunks` return a slice (renamed `appendSeriesChunks`).
* Rename `oooMergedChunks` to `mergedChunks`.
* Improve comment on `ChunkOrIterableWithCopy`.
* Name return values from unpackHeadChunkRef.
Co-authored-by: Oleg Zaytsev <mail@olegzaytsev.com>
Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
Use headIndexReader instead.
OOOCompactionHeadIndexReader needs to be expanded slightly, because it previously delegated to OOOHeadIndexReader.
Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
Just query via `HeadAndOOOQuerier`, which will skip series where no
in-order chunks are in range.
Now we don't need `OOORangeHead`.
Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
Add `HeadAndOOOQuerier` which iterates just once over series, then
where necessary merges chunks from in-order and out-of-order lists.
Add a ChunkQuerier for in-order and ooo together
Add copy-last-chunk behaviour to HeadAndOOOChunkReader
Out-of-order chunk IDs are distinguished from in-order by setting bit 23.
Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
Refactor existing BenchmarkQuerierSelect to provide the set-up.
Note that Head queries now run faster because they use a RangeHead.
Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
Rename a variable.
Add parameters to memSeries.insert function.
No effect on how float samples are handled.
Related to #14546
Signed-off-by: György Krajcsovits <gyorgy.krajcsovits@grafana.com>
Actions have several publish steps that are gated by pushes to tags
starting with v2. This adds the same for tags starting with v3.
Signed-off-by: Jan Fajerski <jfajersk@redhat.com>
This test keeps timing out on our arm64 CI server, it does use a very slow timeout and that 5ms doesn't seem to be enough.
But it 10x.
Signed-off-by: Lukasz Mierzwa <lukasz@cloudflare.com>
Histogram quantile returns NaN in this case, which might be
surprising, so add a unit test that clarifies that this is
intentional.
Signed-off-by: György Krajcsovits <gyorgy.krajcsovits@grafana.com>
Bryan Boreham