Add ReadChangeStream IO param to adjust backlog estimates for replication delay #30995
Conversation
|
@tonytanger for Bigtable review |
jackdingilian
force-pushed
the
autoscaler-replication
branch
from
af5e147
to
36a9809
Compare
|
Checks are failing. Will not request review until checks are succeeding. If you'd like to override that behavior, comment |
jackdingilian
force-pushed
the
autoscaler-replication
branch
from
36a9809
to
d9182c3
Compare
|
Run Spotless PreCommit |
|
LGTM for bigtable review |
|
Assigning reviewers. If you would like to opt out of this review, comment R: @m-trieu for label java. Available commands:
The PR bot will only process comments in the main thread (not review comments). |
...a/io/google-cloud-platform/src/main/java/org/apache/beam/sdk/io/gcp/bigtable/BigtableIO.java
Outdated
Show resolved
Hide resolved
jackdingilian
force-pushed
the
autoscaler-replication
branch
from
d9182c3
to
ee8ebd3
Compare
| @@ -2075,6 +2078,8 @@ static ReadChangeStream create() { | |||
|
|
|||
| abstract @Nullable Boolean getCreateOrUpdateMetadataTable(); | |||
|
|
|||
| abstract @Nullable Duration getBacklogReplicationAdjustment(); | |||
There was a problem hiding this comment.
Can this be Optional autovalue should support this
There was a problem hiding this comment.
I'd prefer to leave this as is for consistency with other optional parameters. We already do a lot of defaulting in ReadChangeStreamIO.expand instead of pushing down optional params into the DoFns and setting defaults there.
I don't have a strong preference for either approach but I think we should keep this consistent with the other parameters.
...platform/src/main/java/org/apache/beam/sdk/io/gcp/bigtable/changestreams/dofn/JodaClock.java
Outdated
Show resolved
Hide resolved
...va/org/apache/beam/sdk/io/gcp/bigtable/changestreams/dofn/ReadChangeStreamPartitionDoFn.java
Outdated
Show resolved
Hide resolved
| @@ -2312,6 +2337,10 @@ public PCollection<KV<ByteString, ChangeStreamMutation>> expand(PBegin input) { | |||
| if (getCreateOrUpdateMetadataTable() != null) { | |||
| shouldCreateOrUpdateMetadataTable = getCreateOrUpdateMetadataTable(); | |||
| } | |||
| Duration backlogReplicationAdjustment = getBacklogReplicationAdjustment(); | |||
There was a problem hiding this comment.
Nit: if using Optional return type can just be
'= getBacklogReplicationAdjustment().orElse(DEFAULT_BACKLOG_REPLICATION_ADJUSTMENT);
There was a problem hiding this comment.
nullable is fine. For streaming IO a consideration is upgrade compatibility, and newly added parameters then need to be nullable and set default in pipeline submission
jackdingilian
force-pushed
the
autoscaler-replication
branch
2 times, most recently
from
07aedb4
to
19a55e2
Compare
|
R: @Abacn for final approval |
jackdingilian
force-pushed
the
autoscaler-replication
branch
from
19a55e2
to
0130bf7
Compare
ReadChangeStreamIO estimates the backlog (via
getSize) for each partition usingpartitionLowWatermark * partitionThroughtput. The low watermark is held back by replication delay for replicated bigtable instances, which can cause the backlog to stay above the downscaling threshold even when the pipeline is 'caught up'This adds a parameter to adjust the backlog downwards based on an estimate of the replication delay. We default it to 30 seconds which testing has shown works well for replicated instances.
Thank you for your contribution! Follow this checklist to help us incorporate your contribution quickly and easily:
addresses #123), if applicable. This will automatically add a link to the pull request in the issue. If you would like the issue to automatically close on merging the pull request, commentfixes #<ISSUE NUMBER>instead.CHANGES.mdwith noteworthy changes.See the Contributor Guide for more tips on how to make review process smoother.
To check the build health, please visit https://github.com/apache/beam/blob/master/.test-infra/BUILD_STATUS.md
GitHub Actions Tests Status (on master branch)
See CI.md for more information about GitHub Actions CI or the workflows README to see a list of phrases to trigger workflows.