So long as they specify timestamps. We don't check that the timestamps
are different.
Extend test, and use client_golang/prometheus/testutil to simplify metric check.
Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
chunkenc.Iterator.AtFloatHistogram may do a shallow copy if
it receives nil as input pointer. This can in turn share the
span slice with multiple histograms in the matrixSelectorHPool,
leading to unexpected errors.
Signed-off-by: György Krajcsovits <gyorgy.krajcsovits@grafana.com>
The histogram stats iterator does not fully clear the histogram object
and is not resilient to new fields being added to the histogram type.
To resolve the issue, the commit uses the CopyTo methods which should
be future proof to new fields being added.
Signed-off-by: Filip Petkovski <filip.petkovsky@gmail.com>
The histogram stats decoder keeps track of the last seen histogram sample
in order to properly detect counter resets. We are seeing an issue where
a histogram with UnknownResetHint gets treated as a counter reset when it follows
a stale histogram sample.
I believe that this is incorrect since stale samples should be completely ignored
in PromQL. As a result, they should not be stored in the histogram stats iterator
and the counter reset detection needs to be done against the last non-stale sample.
Signed-off-by: Filip Petkovski <filip.petkovsky@gmail.com>
[BUGFIX] Storage: errors from a single secondary querier should be warnings.
This is a backport of #14496 to release-2.54 branch.
#13434 introduced an unwanted change in behaviour: if there was no primary querier and a single secondary querier, the secondary would be treated like a primary. This PR restores the previous behaviour, that all secondary queriers report errors as warnings.
In order to test this behaviour, I changed `TestMergeQuerierWithSecondaries_ErrorHandling` so it now calls `NewMergeQuerier` rather than creating the internal data structure directly.
This in turn required all the data types to change, so I merged `mockGenericQuerier` into `mockQuerier`.
Also replaced `unwrapMockGenericQuerier` with a visitor pattern.
While I was there, I addressed the comment from https://github.com/prometheus/prometheus/pull/13434#pullrequestreview-2191058921 to short-circuit the merge of single querier with any number of no-op or nil queriers.
Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
Previously, the config doc suggested that scrapes are simply failed if
the bucket limit is exceeded. However, instead Prometheus first
attempts to reduce the resolution. This will succeed in almost all
non-pathological cases. However, in the rare case that it doesn't, the
scrape will be failed after all.
For the resolution limit AKA `native_histogram_min_bucket_factor`, the
logic is a bit different. Once the highest factor, i.e. schema -4, is
reached, the scrape will still succeed.
Signed-off-by: beorn7 <beorn@grafana.com>
In each case the following member requires 8-byte alignment, so moving
one beside the other shrinks memSeries from 176 to 168 bytes, when
compiled with `-tags stringlabels`.
Signed-off-by: Bryan Boreham <bjboreham@gmail.com>
Bryan Boreham