We have a small database we use to track workorders. It has suddenly
quit responding to certain queries, and I'm not sure how to trace this.
It's running postgresql-7.1-1 on a redhat 6.2 (I know, time to update)
box. Queries outside of coldfusion work fine, but from within the cf app
they just hang and jump the server load to over 4.00.
At this point it smells like an odbc driver problem, but not quite sure
how to debug it. I've got postgres syslog turned on, but nothing shows
up in the log. CF server logs show nothing that sticks out either.
I've read some postings elsewhere re utf-8 being a potential issue, but
no luck on that end so far.
Can someone point me as to where to luck so I can resolve this?
Thanks
Does select * from pg_stat_activity show any connections from CF?
If so, what queries are they running?
Jochem
--
Jochem van Dieten
Team Macromedia Volunteer for ColdFusion, beer and fun.
Please excuse my ignorance - but what database should I connect to in
order to run that query?
Most are select * from <db_name>. I kinda suspect some bad data, but
not sure what to do about it. Tailing logs doesn't really show anything.
Doesn't matter. It shows all current connections to the database,
and if you are a superuser and you have turned on logging you can
even see which queries are being executed.
> Most are select * from <db_name>. I kinda suspect some bad data, but
> not sure what to do about it.
Do you get errors when you do a VACUUM FULL ANALYZE?
AlImage
MX and 7 can use JDBC, 5 can not.
SELECT * from wor_order where <condition>
The query times out...
however, if I do the following:
SELECT (full field list) from wor_order where <condition>
no problem. The SELECT * problem only seems to exist from within CF - I
can use PGADMIN and SELECT * all day long with no problems...
Strange...
That is probably an ODBC problem. I have had similar problems
with the earlier ODBC drivers and cfqueryparam, so I assume it is
something in the area of not being able to identify the datatypes
or column names.
I'm afraid that if updating your ODBC driver doesn't fix it
writing the full field list is your only other option. If you
have any plans for moving to MX or 7, now is the time. I have
never had any such issues with the JDBC drivers.