[Pgpool-general] PgPool 2.2.3 backend stuck in infinite deallocate loop?

[Glyn Astill]

Hi Chaps,

I've just had an odd occourance using 2.2.3. I had a pgpool backend that appeared to be stuck in an infinite loop sending dealocate statements to the postgres server, it was iniing the load avg up to 30 and I had to kill -9 the pgpool backend to stop it.

The client was long gone so I assumed they either crashed out, or this happened as part of the reset_query_list behaviour. 

My postgres log accumulated 2Gb of the following error over a period of 30 minutes or so:

2009-08-26 16:09:03 BST [9590]: [36-1]: [user=nicgicl]: [host=10.10.10.100]: [db=SEE]:: STATEMENT:  DEALLOCATE "find_gt_f3_byindx_6_3"
2009-08-26 16:09:03 BST [9590]: [37-1]: [user=nicgicl]: [host=10.10.10.100]: [db=SEE]:: ERROR:  prepared statement "find_gt_f3_byindx_6_3" does not exist

We've been on 2.2.3 sunce last wednesday, and along with the move I've changed reset_query_list from 'ABORT; RESET ALL; SET SESSION AUTHORIZATION DEFAULT' to 'ABORT; DISCARD ALL'.

This is the first problem I've noticed in a week of operation, however for now I'm switching reset_query_list back.

Any ideas?