Rajesh Kumar Mallah <mallah@trade-india.com> writes:
> ok, you mean we should put statement timeout with statements
> that potentially lock table exclusively. eg
Actually I think it'd work better to put NOLOCK on the read-only
operations. Those guys should never fail to get the lock they need
under ordinary circumstances. If you happen to be running some kind
of schema-altering process in parallel, then the read-only guys will
fail immediately instead of waiting, but AFAICT that's what you wanted.
regards, tom lane