Re: Request: pg_cancel_backend variant that handles 'idle in transaction' sessions

On 11/4/15 4:55 PM, Stephen Frost wrote:
> * Joe Conway (mail@joeconway.com) wrote:
>> On 11/04/2015 01:24 PM, Alvaro Herrera wrote:
>>> I agree with Pavel.  Having a transaction timeout just does not make any
>>> sense.  I can see absolutely no use for it.  An idle-in-transaction
>>> timeout, on the other hand, is very useful.
>>
>> +1 -- agreed
>
> I'm not sure of that.  I can certainly see a use for transaction
> timeouts- after all, they hold locks and can be very disruptive in the
> long run.  Further, there are cases where a transaction is normally very
> fast and in a corner case it becomes extremely slow and disruptive to
> the rest of the system.  In those cases, having a timeout for it is
> valuable.
>
> David (adding him to the CC) actually developed a utility specifically
> to identify what transactions are blocking what others and to kill off
> other processes if they were running for too long and blocking higher
> priority processes.  It didn't matter, in that environment, if they were
> idle-in-transaction or actively running.

You are remembering correctly, Stephen, though there were different
timeouts for blocking transactions that were running and those that were
idle-in-transaction.  We usually set the idle-in-transaction timeout
much lower as it measured not total transaction time but idle time since
the last state change.  In that environment, at least, an
idle-in-transaction session was always due to a stuck process, bug, or
user session left open overnight.  Because partitions and FKs were
continuously being created even ACCESS SHARE locks could be a problem.

The important thing about this implementation was that nothing was
terminated unless it had exceed a timeout AND was blocking another
process.  A feature of this particular system was that it had very long
running transactions that needed to execute unless there was a conflict.

Even then, we'd get an alert some time in advance of the transaction
being terminated so we could make the judgement call to terminate the
other process(es) instead.

--
-David
david@pgmasters.net