Re: PostgeSQL problem (server crashed?)

Edwin Pauli <edwin(at)epauli(dot)dyndns(dot)org> writes:
> My PostgreSQL server is not reachable at this moment. It look likes a
> crash from the postmaster proces.

There is nothing in what you report to indicate that there is anything
wrong with the server at all. It looks to me like you have client
programs that are issuing bogus queries, but you need to go fix that
on the client side; it's hardly the server's fault.

> Warning: pg_query(): Query failed: ERROR: syntax error at or near
> "'5a55385afe83062516f465d5dbb23e96'" at character 64 . in
> /home/epauli/WWW/qst-new/_inc/main.inc.php on line 9

Syntax error in client-generated query.

> Warning: pg_query(): Query failed: ERROR: syntax error at or near
> "hoofdid" at character 8 . in
> /home/epauli/WWW/qst-new/_inc/main.inc.php on line 81

Ditto.

> Warning: pg_fetch_object(): supplied argument is not a valid
> PostgreSQL result resource in
> /home/epauli/WWW/qst-new/_inc/main.inc.php on line 89

Internal problem in PHP (possibly an indication of code that isn't
checking for query failure).

> When i see this problem the first time, i've stopped the postmaster
> proces directly.
> After that, i trying to start the postmaster proces with pg_ctl.

Unnecessary intervention.

> There is only one message logged. That is in /var/log/messages

> Mar 14 00:09:40 epauli postgres[27088]: [1-1] FATAL: the database
> system is starting up
> Mar 14 00:09:42 epauli postgres[27093]: [1-1] ERROR: syntax error at
> or near "d" at character 8
> Mar 14 00:09:43 epauli postgres[27097]: [1-1] ERROR: syntax error at
> or near "d" at character 8
> Mar 14 00:09:44 epauli postgres[27101]: [1-1] ERROR: syntax error at
> or near "d" at character 8
> This lines are repeated 60 times (1 minute)

You've got some client process that is connecting and issuing a bogus
query once a minute. The first of these messages occurred because it
tried to do so while the database was still restarting. However, the
restart clearly completed without incident, because we see fresh backend
sessions running to issue the later messages.

> Does anybody know what's wrong?

Nothing on the postgres side, AFAICS. Possibly you need a PHP guru.
You might also try turning on log_statement so you can see the full text
of the invalid SQL commands in the log; that might help track down the
source.

regards, tom lane