Many bad alloc errors, dispatch directory fills up and does not free up space for other jobs

[Zarack]

I have a SH cluster with 3 servers, but I'm getting a lot of replication errors because the datamodels fill up the dispatch directory.
How are jobs released from dispatch? Are files cleaned automatically?

There are many bad alloc errors.

Zarack_1-1699465442059.png

Thanks.

[richgalloway]

Jobs should be removed from the dispatch directory automatically once they expire.

Some workarounds to consider:

• Increase the size of the dispatch directory
• Reduce how frequently datamodel accelerations run
• Disable unused DMAs