kamilwu commented on a change in pull request #12542:
URL: https://github.com/apache/beam/pull/12542#discussion_r471557407
##########
File path: .test-infra/jenkins/job_LoadTests_coGBK_Python.groovy
##########
@@ -147,25 +147,30 @@ def loadTestConfigurations = { datasetName ->
autoscaling_algorithm: 'NONE'
]
],
- ].each { test -> test.pipelineOptions.putAll(additionalPipelineArgs) }
+ ]
+ .each { test -> test.pipelineOptions.putAll(additionalPipelineArgs) }
+ .each { test -> (mode) != 'streaming' ?: addStreamingOptions(test) }
}
-def batchLoadTestJob = { scope, triggeringContext ->
- scope.description('Runs Python CoGBK load tests on Dataflow runner in batch
mode')
- commonJobProperties.setTopLevelMainJobProperties(scope, 'master', 240)
+def addStreamingOptions(test) {
+ // Use highmem workers to prevent out of memory issues.
+ test.pipelineOptions << [streaming: null,
+ worker_machine_type: 'n1-highmem-4'
Review comment:
My pipelines kept crushing because of OutOfMemory exceptions. So I
followed the advice given in this article:
https://cloud.google.com/community/tutorials/dataflow-debug-oom-conditions
Other solution was to, according to the article, use fewer number of threads
per worker. But I couldn't have found a pipeline option responsible for that in
Python SDK (it exists in Java SDK, though).
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
