The series index looks at a set of tombstones when querying the id for
a given key, but it does not look when asking for the offset for some
id, even if that id is deleted.
Update the verify tooling to check that the index agrees with the
deleted status of the id, but skip doing the extra checks if the
id is deleted.
The reduce iterators would read in the points for a window, which
matched the grouping of the outermost query, and then it would sort them
by the time before emitting the points.
When there were multiple series, this would sometimes cause a conflict
because it would change the sorting of the inner query output when
selectors were used within a subquery. Then, these emitted points would
be output in the wrong order and they wouldn't join correctly when
multiple cursors were used.
This fixes it so the sorting happens per series grouping rather than on
all of the points together so they retain their tag order which is the
correct sorting method.
When a Flux predicate is transformed to a Store.Read / GroupRead request,
the `_measurement` and `_field` keys are remapped to match 2.x internal
tag keys.
This change does not modify the 2.x behavior, but rather updates the
1.x mapping, so merging future updates from the 2.x storage/reads
package should have fewer conflicts.
This integrates the influxdb 1.x series to the latest version of Flux
and updates the code to use it. It also removes the dependency on
platform and copies the necessary code from storage into the 1.x series
so the dependency is unneeded.
The flux functions specific to 1.x have been moved to the same structure
that flux changed to with having a `stdlib` directory instead of a
`functions` directory. It also adds a `databases()` function that
returns the databases from the meta client.
Previously it was possible to set IDs on a `nil` entry which would
in turn cause a panic. If this panic was recovered by the server
then it would result in a mutex in the `inmem` index staying locked
indefinitely.
This commit extends the Prometheus remote write endpoint to drop
unsupported Prometheus values, rather than reject the entire batch.
InfluxDB does not support NaN, -Inf or +Inf, but Prometheus does. The
remote write endpoint will now drop these and write valid values in the
provided batch.
If the user enabled write trace logging (`[http] write-tracing = true`)
then summaries of any dropped values within a batch will be logged.
If a batch of values contains any values that are subsequently dropped,
the returned status code will be `204`.
Specifically:
* renamed files for consistency between versions
* added `time-interval` schema option
* updated schema example documentation
Back port of improvements from #12710
We're not allowed to access the s.epochs map without holding the
mutex against shard creation and deletion, so create a copy of
all of the epoch trackers we will need while we hold the mutex.
Lorenzo Affetti
Hercules Mango Churchill
Adam
Adam
Jonathan A. Sternberg
Jeff Wendling
Jeff Wendling
Stuart Carnie
Ben Johnson
David Norton
Edd Robinson
Ahmed Ashour