The following appeared in the log this am, while running "vacuumdb":
[13323] DEBUG: ReceiveSharedInvalidMessages: cache state reset
[25461] DEBUG: SendSharedInvalidMessage: SI buffer overflow
[13322] DEBUG: ReceiveSharedInvalidMessages: cache state reset
[13323] DEBUG: ReceiveSharedInvalidMessages: cache state reset
[25461] DEBUG: SendSharedInvalidMessage: SI buffer overflow
[13322] DEBUG: ReceiveSharedInvalidMessages: cache state reset
[25461] DEBUG: ReceiveSharedInvalidMessages: cache state reset
Server version is PostgreSQL 7.2.2 on i686-pc-linux-gnu, compiled by GCC
3.0.4.
vacuumdb was run in a PuTTY session.
The server is a quad-processor Xeon with 6Gb of Micron memory (qualified by
Micron for the server) and a Mylex Acceleraid card (replaced because of
flaky behavior causing "signal 11" restarts of the postmaster, which no
longer occur); in other words, the hardware is sound, and the software is at
its latest version. (I did see this error once or twice several weeks ago,
but that was before we replaced the problem RAID controller, so I put off
worrying about it till we replaced the bad hardware. This is the first time
since that the errors have occurred. The server intermittently sees heavy
loads - ad hoc queries on )
The server runs linux-xfs v1.1 compiled with gcc 2.96.
Some postgreql.conf lines relating to shared memory:
shared_buffers = 131072
wal_buffers = 64
sort_mem = 512
vacuum_mem = 8192
Any ideas on the source of these errors and the implications for data
corruption?
Thanks,
Murthy