Writing points that were not sorted by time could cause very high
CPU usages and increased latencies because each point inserted would
cause the in-memory cache to be resorted. The worst case would be
writing a large batch of N points in reverse time order which would
invoke N sorts of the slice.
This patch keeps track of which slices need to be sorted and sorts
them once at the end. In the previous example, the N sorts becomes
one. There is still a pathalogical case that would require N/2 sorts.
For example, 10000 points split across 5000 series. Each series has two
points that are in reverse time order. This would incur 5000 sorts still.
Fixes#3159
This commit adds a write ahead log to the shard. Entries are cached
in memory and periodically flushed back into the index. The WAL and
the cache are both partitioned into buckets so that flushing doesn't
stop the world as long.
Fixes#2869
When adding a new field to an existing measurment, Shard.validateSeriesAndFields
would also encode the fields as a side-effect. In the case of a new field
that needed to be created, the encoding would fail because the field type
had not been created for the measurement yet. The fields are re-encoded
after validateSeriesAndFields returns and after the field encoding have been
setup properly so this additional encoding during
validation isn't necessary.
* Add deleteMeasurement to store and shard
* Add DropMeasurement to DatabaseIndex
* Update ErrMeasurementNotFound and ErrDatabaseNotFound to not include the first line of the stack trace.
* Pulled over updates to ast and parser from master
* Updated store and shard to be able to drop series
* Pulled updates to database.go from master into tsdb/meta.go
Fixes issue where queries wouldn't be able to hit anything because the index does't load until the shard is open.
Fix an issue where field codecs weren't populated in the shard when loading.
This allows the new write path to be hooked up if you start the
server with `INFLUXDB_ALPHA1=1`. When set, writes will go though
the coordinator and be stubbed out to write to a single local data
node with one shards. The write will be logged and written to
disk .
The env var is used so that the current write path is not completely
broken which would break many of the tests that depend on writes.
Note that queries are not currently working w/ the this change.
Ben Johnson
Alexander Morozov
Jason Wilder
Philip O'Toole
Joseph Crail
Paul Dix
Philip O'Toole
David Norton
Cory LaNou