Closed jdposthuma closed 3 years ago
Oh boy.
I have added support for streaming bulk index mode. This is however not the default. Streaming bulk is more lenient on memory usage. You can enable it by defining
ELASTICSEARCH_STREAMING_BULK=True
This was wholly caused by the issues surrounding #164 and #162. Because those each provide more specific root causes, I suggest this be closed.
ELASTICSEARCH_STREAMING_BULK=True
did not positively affect performance.
Closing this
PGSync version: 2.1.1
Postgres version: 12.x
Elasticsearch version: 7.10
Redis version: 6.0
Python version: 3.7, 3.8
Problem Description: When syncing a table (root has ~1.3 million records, child has 10 records) in daemon mode (
pgsync -d
). It works initially, but then bogs down and is killed with exit code 137 (out of memory). I've been able to replicate this on a container with 4GB of memory.How to avoid this?
Error Message (if any):