Hi all;
we have a client experiencing the following scenario:
- Postgresql 8.3.10 plus postgis 1.5.1
- Both databases stop responding to queries and several of the backend
processes show as being in an uninterruptible state (state "Us" according to
a ps aux)
- The client killed the processes, did a pg_dumpall and a restore only to find
the condition happen again the next day
- The client read every file reported by lsof as being accessed by a backend
process without issues.. thus it seems to NOT be a disk I/O issue.
Thoughts?