Golden Gate Replicat Consuming Large Amount Of Memory And Growing Linux x86-64 - Teradata
Last updated on OCTOBER 10, 2016
Applies to:Oracle GoldenGate - Version 126.96.36.199.1 and later
Information in this document applies to any platform.
While running replicats with BATCHSQL and HANDLECOLLISIONS (processing trail files generated during initial load process), the replicts are reporting: WARNING OGG-01137 BATCHSQL suspended, continuing in normal mode. In normal mode, with handlecollisions, the collision is then handled and the replicat returns to BATCHSQL mode: INFO OGG-01139 BATCHSQL resumed, recovered from error.
When we see this occur over and over, we see the memory consumption of the server for the replicat continue to grow.
example, using ps to capture memory consumption:
this replict has been up since (per rpt file:) Starting at 2015-04-10 13:34:09
793.133 MB RNSC3
Fri Apr 10 13:37:22 EDT 2015
2087.46 MB RNSC3
Fri Apr 10 13:42:22 EDT 2015
2985.18 MB RNSC3
Fri Apr 10 13:47:22 EDT 2015
3912.5 MB RNSC3
Fri Apr 10 13:52:22 EDT 2015
If we remove BATCHSQL param, the memory consumption remains below 100MB for over 12 hours.
We see this behavior in multiple replicats.
GoldenGate for Teradata 188.8.131.52
If we remove BATCHSQL, so as not to consume memory, the replicats are too slow and lag grows.
For replicats that are current and no longer require HANDLECOLLISIONS, we don't see this growth when using BATCHSQL.
We only see it in replicts with BATCHSQL that switch to normal mode and return to batch sql mode.
Sign In with your My Oracle Support account
Don't have a My Oracle Support account? Click to get started
My Oracle Support provides customers with access to over a
Million Knowledge Articles and hundreds of Community platforms