When using queries like 'select count(_seriesKey) from bigmeasurement`, we
should iterate over the tsi structures to serve the query instead of loading
all the series into memory up front.
Co-authored-by: Sam Arnold <sarnold@influxdata.com>
When the compaction planner runs, if it cannot acquire
a lock on the files it plans to compact, it returns a
nil list of compaction groups. This, in turn, sets the
engine statistics for compactions queues to zero,
which is incorrect. Instead, use the length of pending
files which would have been returned.
closes https://github.com/influxdata/influxdb/issues/22138
(cherry picked from commit 7d3efe1e9e)
closes https://github.com/influxdata/influxdb/issues/22141
This fix ensures that memory-mapped files are not released
before pointers into them are copied into heap memory.
MeasurementNamesByExpr() and MeasurementNamesByPredicate() can
cause panics by copying memory from mmapped files that have been
released. The functions they call use iterators to files which
are closed (releasing the mmapped files) before the memory is
safely copied to the heap.
closes https://github.com/influxdata/influxdb/issues/22000
(cherry picked from commit a989f8f8b6)
closes https://github.com/influxdata/influxdb/issues/22002
tsdb.Engine.IsIdle and tsdb.Engine.Digest now return a reason string for why the engine & shard are not idle.
Callers can then use this string for logging, if desired. The returned reason does not allocate memory, so the
caller may want to add the shard ID and path for more information in the log. This is intended to be used in
calls from the anti-entropy service in Enterprise.
(cherry picked from commit bf45841359)
fixes https://github.com/influxdata/influxdb/issues/21448
(cherry picked from commit c8da9bafbf)
closes https://github.com/influxdata/influxdb/issues/21894
Under heavy write load creating new fields and measurements
the rewrite of the fields.idx file is a bottleneck. This
enhancement combines multiple writes into a single one and
shares any error return value with all of the combined
invocations. MeasurementFieldSet and the new
MeasurementFieldSetWriter must both now be explicitly
closed.
Closes#21577
(cherry picked from commit f64be286be)
Closes https://github.com/influxdata/influxdb/issues/21598
The cursors were previously [start, stop) to be consistent with how flux
requests data, but the underlying storage file store was [start, stop]
because that's how influxql read data. This reverts back the cursor
behavior so that it is now [start, stop] everywhere and the conversion
from [start, stop) to [start, stop] is performed when doing the cursor
request to get the next cursor.
Co-authored-by: Sam Arnold <sarnold@influxdata.com>
fields.idx frequent writes cause lock contention and fields.idx is recreated
when a field or measurement is added in a WritePointsWithContext()
This eliminates locking during the actual file rewrite, and limits it to
the times when the MeasurementFieldSet is actually being read or written
in memory and when the new file is being renamed.
Test verification of correct behavior by checking the fields.idx
file matches the in-memory copy after heavily parallel measurement addition.
Co-authored-by: davidby-influx <72418212+davidby-influx@users.noreply.github.com>
This use-after-free bug may lead to segfault. The iterators that have
reference to the underlying index files were closed too early while
the bitmaps were still used afterwards. If a compaction occurs
concurrently and removes the index files, it would result in accessing
unmap'd memory address.
This is a backport of #14262 to the 1.x storage engine. The 1.x storage
engine is now the primary engine for open source so when we switched we
regressed to the old behavior.
This also fixes `go generate` for the tsm1 package by running `tmpl`
with `go run` instead of assuming the correct one is installed in the
path.
This fixes multi measurement queries that go through the storage service
to correctly pick up all series that apply with the filter. Previously,
negative queries such as `!=`, `!~`, and predicates attempting to match
empty tags did not work correctly with the storage service when multiple
measurements or `OR` conditions were included.
This was because these predicates would be categorized as "multiple
measurements" and then it would attempt to use the field keys iterator
to find the fields for each measurement. The meta queries for these did
not correctly account for negative equality operators or empty tags when
finding appropriate measurements and those could not be changed because
it would cause a breaking change to influxql too.
This modifies the storage service to use new methods that correctly
account for the above situations rather than the field keys iterator.
Some queries that appeared to be single measurement queries also get
considered as multiple measurement queries. Any query with an `OR`
condition will be considered a multiple measurement query.
This bug did not apply to single measurement queries where one
measurement was selected and all of the logical operators were `AND`
values. This is because it used a different code path that correctly
handled these situations.
* refactor: Replace ctx.Done() with ctx.Err()
Prior to this commit we checked for context cancellation with a select
block and context.Context.Done() without multiplexing over any other
channel like:
select {
case <-ctx.Done():
// handle cancellation
default:
// fallthrough
}
This commit replaces those type of blocks with a simple check of
ctx.Err(). This has the following benefits:
* Calling ctx.Err() is much faster than entering a select block.
* ctx.Done() allocates a channel when called for the first time.
* Testing the result of ctx.Err() is a reliable way of determininging if
a context.Context value has been canceled.
* fix: Fix data race in execDeleteTagValueEntry()
This commit adds `mincore.Limiter` which throttles page faults caused
by mmap() data. It works by periodically calling `mincore()` to determine
which pages are not resident in memory and using `rate.Limiter` to
throttle accessing using a token bucket algorithm.
* chore: remove tsi1 testdata and add go generate file to download
* chore: fix testdata url and rename gen file
* fix: add testdata generate command to Makefile
* chore: add testdata dir to gitignore
* refactor(tsdb): improve error message when missing testdata
* refactor(tsdb): tagged testdata and avoid stacktrace when missing
Checking a channel too regularly could cause
context switching to other goroutines. In tight loops,
it is prudent to check, but to do so less frequently so
as to avoid thrashing.
* feat(tsdb): SHOW TAG KEYS (no time) query using only TSI data.
* fix(tsdb): Allow for earlier return when scanning during show tag keys.
* fix(tsdb): Speed things up by using the key merger to reduce allocs.
* chore(tsm1): Fix golint.
* fix(tsdb): Remove sorting, because these keys should already be sorted.
* fix(tsdb): Remove dead code to placate the linter.
* fix(tsm1): delimit tsmKeyPrefix with appended comma
Fixes#7589.
Append a comma to the TSM key prefix when matching a full measurement name to avoid erroneously matching other measurement names that include the prefix in their own name. For example, this prevents matching a measurement "cpu1" when targeting "cpu" by updating the prefix to "cpu,". This relies on the fact that tag key-value pairs are separated by commas.
* fix(tsm1): regression tests for tsmKeyPrefix comma delimiting
* fix(storage): Push-down a predicate to match tags for SHOW MEASUREMENTS calls.
* chore: Address feedback.
* fix(tsm1): Split behavior based on existence of predicate for show measurements.
* fix(tsm1): Allow parenthesis expression on the LHS of a predicate.
* fix(tsm1): Create a separate tag predicate verifier that rejects negative comparisons.
* fix(tsm1): Additional test cases for show measurements with predicate.
This commit adds ref counting for files that we pull tag keys from.
Previously, files were only ref counted during the time we extracted
tag keys but this commit adds additional ref counting for the life of
the `Engine.tagKeysNoPredicate()` function.
This commit
* adds new request and response data types for schema gRPC calls
* adds fmt.Stringer implementation to cursors.FieldType
* adds APIs to sort a slice of MeasurementField values,
* upgrades the gogo protobuf package to v1.3.1, which
includes improvements to serialization.
This commit adds a new API to `Cache` to address data races
with the `TagKeys` and `TagValues` APIs.
`Cache` and `entry` provide `AppendTimestamps`, which
appends the current timestamps to the provided slice
to reduce allocations. As noted in the documentation,
it is the responsibility of the caller to sort and deduplicate
the values, if required.
The `cursors.TimestampArray` type was extended to permit
use of the `sort.Sort` API.
This commit introduces a new API for finding the maximum
timestamp of a series when iterating over the keys in a
set of TSM files.
This API will be used to determine the field type of a single
field key by selecting the series with the maximum timestamp.
It has also refactored the common functionality for iterating
TSM keys into `timeRangeBlockReader`, which is shared
between `TimeRangeIterator` and `TimeRangeMaxTimeIterator`.
Daniel Moran
Jonathan A. Sternberg
davidby-influx
Sam Arnold
Dane Strandboge
Yun Zhao
LLThomas
Tristan Su
Ben Johnson
Stuart Carnie
jl
Mark Rushakoff
sans
Roger Peppe
Ayan George
Brett Buddin
Edd Robinson
Gavin Cabbage
ricky
Phil Bracikowski
jlapacik
Patrik Helia