"Laura Hornbeck" <lhornbeck@oppunl.com> writes:
> I am repeatedly getting invalid memory alloc request size errors on almost a
> daily basis and not always on the same table. To fix the troubled table, I
> find the record that either gives the memory error or makes postgres crash
> (gives a terminated by signal 11 error in the log) and delete it. Until I
> do that, I cannot select the data from the table by any means (select * ,
> dump, copy to, etc).
This sounds like nothing so much as flaky hardware. Could be bad RAM,
loose cables, disk drive on the edge of failure ...
You need to take the machine out of service for a bit and run some
hardware diagnostics on it. memtest86 is widely recommended for testing
RAM, and you could use badblocks or something similar to look for
possible disk flakiness. The old "clean, and reseat all the connectors"
drill would be a worthwhile step too.
regards, tom lane