feat: optional dedicated WAL volume for dataservers - #13
Merged
Conversation
Add server.wal.{storage,storageClassName} to give the WAL its own
volumeClaimTemplate, mounted at the configured storage.wal.dir
(default /data/wal), so WAL fsyncs don't compete with database I/O
for the same disk — and on cloud block storage each volume brings its
own IOPS budget. The WAL claim inherits server.storageClassName unless
its own storageClassName is set (e.g. a faster class just for the WAL).
When server.wal is unset the chart renders exactly as before (single
data volume), so existing installs are unaffected. Chart 0.0.8.
Signed-off-by: Matteo Merli <mmerli@apache.org>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds
server.wal.storage(and optionalserver.wal.storageClassName) to the oxia-cluster chart: when set, dataservers get a secondvolumeClaimTemplatenamedwal, mounted at the configuredserver.config.storage.wal.dir(default/data/wal).Why: separating the WAL from the database puts fsync-latency-critical writes on their own device — WAL appends stop competing with pebble compaction/read I/O, and on cloud block storage (e.g. EBS gp3) each volume brings its own IOPS/throughput budget. The
storageClassNameoverride also allows a faster class just for the WAL. This mirrors the standard deployment practice for ZooKeeper (dataLogDir) and etcd (--wal-dir).Compatibility:
server.walunset renders byte-identical to 0.0.7 (singledatavolume) — existing installs unaffected. The WAL claim inheritsserver.storageClassNameunless its own class is set. Mounting at the configured wal dir means the dataserver config needs no change.Chart version bumped to 0.0.8.
Verification
helm lintclean;helm templatechecked in three modes:dataclaim, no wal mount (unchanged)--set server.wal.storage=10Gi --set server.storageClassName=gp3→walclaim (10Gi, gp3) mounted at/data/wal--set server.wal.storageClassName=io2→ wal claim on io2 while data stays on gp3