On 11/04/2013 01:56 PM, Jeff Amiel wrote:
On Monday, November 4, 2013 3:23 PM, Adrian Klaver <adrian.klaver@xxxxxxxxx> wrote:
Probably poor choice of words:). So then, what we are looking at is
other clients trying to update user_profile but not succeeding because
pid 4899 is blocking. At this point all I can see is that the offending
query is updating some fields the others are not; disabled and reset_code.
Is that always the case?
If so any thing in the code path that is different when those fields are
updated?
We have scenarios where exact same query is in play in all instances.
Which query is that?
And what scenario are you talking about, blocking query or something else?
Any thoughts as to the fact that this could be a full table_lock simply based on the use of username (non primary key - but specifically unique constraint) in the where clause? I'm grasping I know....
What makes you think the username condition is the problem?
--
Adrian Klaver
adrian.klaver@xxxxxxxxx
--
Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general
