* test(chunkenc): appending histograms with empty buckets and gaps
Append such native histograms that have empty buckets and gaps
between the indexes of those buckets.
There is a special case for appending counter native histograms to a chunk in TSDB: if we append a histogram that is missing some buckets that are already in chunk, then usually that's a counter reset. However if the missing bucket is empty, meaning its value is 0, then we don't consider it missing.
For this case to trigger , we need to write empty buckets into the chunk. Normally native histograms are compacted when we emit them , so this is very rare and compact make sure that there are no multiple continuous empty buckets with gaps between them.
The code that I've added in #14513 did not take into account that you can bypass compact and write histograms with many empty buckets, with gaps between them. These are still valid, so the code has to account for them.
Main fix in the expandIntSpansAndBuckets and expandFloatSpansAndBuckets function. I've also refactored them for clarity. Consequently needed to fix insert and adjustForInserts to also allow gaps between inserts.
I've added some new test cases (data driven test would be nice here, too many cases). And removed the deprecated old function that didn't deal with empty buckets at all.
Signed-off-by: György Krajcsovits <gyorgy.krajcsovits@grafana.com>
Signed-off-by: George Krajcsovits <krajorama@users.noreply.github.com>
Co-authored-by: Björn Rabenstein <beorn@grafana.com>
Always return unknown hint for first sample in non-gauge histogram chunk
---------
Signed-off-by: Fiona Liao <fiona.liao@grafana.com>
Co-authored-by: György Krajcsovits <gyorgy.krajcsovits@grafana.com>
* chunkenc: allow missing empty buckets on histogram append
Allow appending to chunks when the histogram to be added is missing
some buckets, but the missing buckets are empty in the chunk.
For example bucket at index 5 is present in the chunk, but its value
is 0 and the new histogram doesn't have a bucket at index 5.
This fixes an issue of merging chunks where one chunk was recoded to
retroactively have some empty buckets in all the histograms and we are
merging in a histogram that doesn't have the empty bucket (because it
was not recoded yet).
The operation alters the histogram that is being added, however this has
already been the case when appending gauge histograms. Thus the test
TestHistogramSeriesToChunks in storage package is changed to explicitly
test what happened to the appended histogram - Compact(0) call is removed.
The new expandIntSpansAndBuckets and expandFloatSpansAndBuckets functions
are a merge of expandSpansForward and counterResetInAnyBucket and
counterResetInAnyFloatBucket.
Signed-off-by: György Krajcsovits <gyorgy.krajcsovits@grafana.com>
* Use single bit to differentiate between optimized bounds and floats
Use one bit to decide what kind of data to read/write.
This reduces storage need of floats from 72 bits to 65 bits and makes the
integers store in 5 to 32 bits instead of 16.
Signed-off-by: György Krajcsovits <gyorgy.krajcsovits@grafana.com>
Signed-off-by: Jeanette Tan <jeanette.tan@grafana.com>
Signed-off-by: György Krajcsovits <gyorgy.krajcsovits@grafana.com>
Signed-off-by: Jeanette Tan <jeanette.tan@grafana.com>
Signed-off-by: George Krajcsovits <krajorama@users.noreply.github.com>
Co-authored-by: Jeanette Tan <jeanette.tan@grafana.com>
Case a) empty span is at the beginning of the spans.
Case b) two consequtive empty spans with positive offsets.
Signed-off-by: György Krajcsovits <gyorgy.krajcsovits@grafana.com>
The code did not handle spans with 0 length properly.
Spans with length zero are now skipped in the comparison.
Span index check not done against length-1, since length is a unit32,
thus subtracting 1 leads to 2^32, not -1.
Fixes and unit tests for both integer and float histograms added.
Signed-off-by: György Krajcsovits <gyorgy.krajcsovits@grafana.com>
This is a bit more conservative than we could be. As long as a chunk
isn't the first in a block, we can be pretty sure that the previous
chunk won't disappear. However, the incremental gain of returning
NotCounterReset in these cases is probably very small and might not be
worth the code complications.
Wwith this, we now also pay attention to an explicitly set counter
reset during ingestion. While the case doesn't show up in practice
yet, there could be scenarios where the metric source knows there was
a counter reset even if it might not be visible from the values in the
histogram. It is also useful for testing.
Signed-off-by: beorn7 <beorn@grafana.com>
- Adjust doc comments to go1.19 style.
- Break down some overly long lines.
- Minor doc comment tweaks and fixes.
- Some renaming.
Some rationales for the last point:
I have renamed “interjections” into “inserts”, mostly because it is
shorter, and the word shows up a lot by now (and the concept is
cryptic enough to not obfuscate it even more with abbreviations).
I have also tried to find more descriptive naming for the “compare
spans” functions.
Signed-off-by: beorn7 <beorn@grafana.com>
This is to check if a gauge histogram can be appended to the given chunk.
If not, it tells what changes to make to the chunk and the histogram
if possible.
Signed-off-by: Ganesh Vernekar <ganeshvern@gmail.com>
This guaranees that the zero threshold is byte-aligned. Not sure if
that helps in any way, but at least it won't harm.
Signed-off-by: beorn7 <beorn@grafana.com>
This adds bit buckets for larger numbers to varbit encoding and also
an unsigned version of varbit encoding.
Then, varbit encoding is used for all the histogram chunk data instead
of varint.
Signed-off-by: beorn7 <beorn@grafana.com>
A lot of this code was hacked together, literally during a
hackathon. This commit intends not to change the code substantially,
but just make the code obey the usual style practices.
A (possibly incomplete) list of areas:
* Generally address linter warnings.
* The `pgk` directory is deprecated as per dev-summit. No new packages should
be added to it. I moved the new `pkg/histogram` package to `model`
anticipating what's proposed in #9478.
* Make the naming of the Sparse Histogram more consistent. Including
abbreviations, there were just too many names for it: SparseHistogram,
Histogram, Histo, hist, his, shs, h. The idea is to call it "Histogram" in
general. Only add "Sparse" if it is needed to avoid confusion with
conventional Histograms (which is rare because the TSDB really has no notion
of conventional Histograms). Use abbreviations only in local scope, and then
really abbreviate (not just removing three out of seven letters like in
"Histo"). This is in the spirit of
https://github.com/golang/go/wiki/CodeReviewComments#variable-names
* Several other minor name changes.
* A lot of formatting of doc comments. For one, following
https://github.com/golang/go/wiki/CodeReviewComments#comment-sentences
, but also layout question, anticipating how things will look like
when rendered by `godoc` (even where `godoc` doesn't render them
right now because they are for unexported types or not a doc comment
at all but just a normal code comment - consistency is queen!).
* Re-enabled `TestQueryLog` and `TestEndopints` (they pass now,
leaving them disabled was presumably an oversight).
* Bucket iterator for histogram.Histogram is now created with a
method.
* HistogramChunk.iterator now allows iterator recycling. (I think
@dieterbe only commented it out because he was confused by the
question in the comment.)
* HistogramAppender.Append panics now because we decided to treat
staleness marker differently.
Signed-off-by: beorn7 <beorn@grafana.com>
