Joe Maldonado <jmaldonado@webehosting.biz> writes:
> While researching this locking issue I got some of the logs and found
> that in one of the cases there was a SELECT running for a long time,
> about 2 hours. This select statement does not usually take more than a
> few seconds though, it appeared that TRUNCATE was waiting on it to
> finish before continuing.
> The SELECT statement in question contains a sub SELECT in the FROM
> clause which in turn is joining with a view that contains the table
> which TRUNCATE is being executed against.
> Is it possible that the SELECT was issues just before the TRUNCATE
> statement was issues and the view in the sub SELECT was waiting on
> TRUNCATE's lock?
No. That would be a deadlock and would be reported as such.
regards, tom lane