Обсуждение: simple_heap_update problem

For the last few weeks (still haven't figure out what changed before
this started) one of our apps has been generating this error message on
the backend one or two times a day. I'm pretty sure it's not an
application error since it can be successfully re-run with the same
inputs. The application does several queries and I'm not sure which one
causes the error. I searched the web and found this error being caused
by concurrent ANALYZE commands, but that's not happening here.

Aug 7 13:23:30 calchas postgres[27794]: [3] ERROR: simple_heap_update:
tuple concurrently updated Aug 7 13:23:30 calchas postgres[27794]: [4]
WARNING: AbortTransaction and not in in-progress state

Any ideas? I realize this isn't too much information -- what can I do
(besides logging queries, which I just started) to get more?
       --Nate

Oh yea, I'm running 7.3.3 on Redhat 7.3.
       --Nate

nmueller@cs.wisc.edu writes:
> Aug 7 13:23:30 calchas postgres[27794]: [3] ERROR: simple_heap_update:
> tuple concurrently updated

The only case of this I'm aware of is the concurrent-ANALYZE issue,
which is pretty harmless.  But you say it's not that.  I'd suggest
turning on log_statement and log_duration so that you can determine
(a) what query is reporting this failure and (b) what was running
concurrently with it.  When you have that info for a few occurrences,
send it along and we'll try to figure out what's happening.
        regards, tom lane