[cobalt-users] Postgresql Problems with Raqs

[RPenna@xxxxxxxxxxxx]

Hi all,

   We have 12 Raq 3/4 systems. Over the past couple weeks, the postmaster
process on a few of them has started to go a little haywire:

 PID USER     PRI  NI  SIZE  RSS SHARE STAT  LIB %CPU %MEM   TIME COMMAND
29744 postgres  17   0   876  124    76 R       0 92.9  0.0 137:06
postmaster

   After looking around further, there are files in the
/home/pgsql/base/cobalt directory called pg_attribute_relid_attnum_index.1,
pg_attribute_relid_attnum_index.2, etc etc, up to 25,000! They're all zero
byte files. I shut down postgres, removed the files, restarted postgres and
all was normal again for about a day. Then, after about 18 hours, they
re-appeared again.

   From what I've seen around, the internal index has been corrupted and I
need to re-index. However, I'm not sure exactly how to do this. I thought
it was kind of weird it started happening on 3 boxes that we've had for
around the same amount of time. Has anyone else experienced this?

   FYI - The boxes are all patched and up to date with the recommended
cobalt packages. Any help with the re-indexing commands would be
appreciated. Thanks in advance.


Ron Penna
Systems Engineer
Netrica Inc.
phone: 716-340-1969
email: rpenna@xxxxxxxxxxx