Re: BUG #4243: Idle in transaction - Mailing list pgsql-bugs

From Andrew Sullivan
Subject Re: BUG #4243: Idle in transaction
Date
Msg-id 20080617133250.GA68434@commandprompt.com
Whole thread Raw
In response to BUG #4243: Idle in transaction  ("Clemens Wagner" <clemens.wagner@denkwerk.com>)
List pgsql-bugs
On Tue, Jun 17, 2008 at 06:49:57AM +0000, Clemens Wagner wrote:
> I have sometimes hanging psql backend processes with "idle in transaction"

There is no bug here.  Some transaction is stuck waiting for something
else, or else you've left a transaction hanging around that way (is
your JDBC driver old?  It used to issue BEGIN right after COMMIT when
in non-autocommit mode.  That'd cause this issue).

You need to figure out what's blocking what.  Look in the pg_locks
system view.  See the manual for more on this.

> high traffic on the database. It leads sometimes to an inconsisten database
> with double primary keys in a table or a broken unique indexes (
> "pg_statistic_relid_att_index" for instance). I can't enforce this error.

I don't believe this is the problem you're having there.  If what
you're saying is the case, you need to look for breakage elsewhere.

A

--
Andrew Sullivan
ajs@commandprompt.com
+1 503 667 4564 x104
http://www.commandprompt.com/

pgsql-bugs by date:

Previous
From: Tom Lane
Date:
Subject: Re: BUG #4244: ALTER TABLE ... NO INHERIT problem
Next
From: "Kevin"
Date:
Subject: BUG #4245: Product Name...