- I’ve nailed down the particular task at hand causing the OOM to 2999348,
- It is related to T1351,
- I suspect the so-called “Transaction Publish Worker” task executed for it is attempting to tokenize the contents of that (large) task,
- because the contents of the entire task’s description is about 4.2MB,
- I’ve limited the POSIX user under which the service runs to a chunk of memory significantly less than the memory available to the entire system.
Does it stop here? Of course not. We’ll need to learn what it is that causes the daemons to consume as much memory more accurately, go upstream and report it.
That being said, the immediate problems should be over.
Thank you for your patience, and apologies for the inconvenience.