From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Neil Conway <neilc(at)samurai(dot)com> |
Cc: | pgsql-patches(at)postgresql(dot)org |
Subject: | Re: CREATE TRIGGER locking |
Date: | 2006-07-04 13:49:27 |
Message-ID: | 13668.1152020967@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-patches |
Neil Conway <neilc(at)samurai(dot)com> writes:
> Last year, I questioned why CREATE TRIGGER acquires an
> AccessExclusiveLock on its target table:
> http://archives.postgresql.org/pgsql-hackers/2005-03/msg00764.php
> Acquiring an ExclusiveLock should be sufficient: we can safely allow
> concurrent SELECTs on the table.
After re-reading that whole thread, I remain just as uncomfortable as
I was here:
http://archives.postgresql.org/pgsql-hackers/2005-03/msg00819.php
You haven't proposed any use-case at all that justifies taking risks
with insufficient locking of DDL changes. It's too early in the morning
for me to think of a specific counterexample, but I think the general
line of "TX A starts to work with a table, and then the schema changes
while it's still active" is relevant. Another problem is that a
transaction that issues CREATE TRIGGER and then some other
schema-changing operation on the same table would likely find itself
trying to upgrade lock from ExclusiveLock to AccessExclusiveLock,
with a very strong chance of deadlock. This latter scenario seems more
probable to me than the case of "I need to add a trigger while allowing
SELECTs to proceed".
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Joe Conway | 2006-07-04 17:40:30 | kerberos related warning |
Previous Message | Bruce Momjian | 2006-07-04 13:23:01 | Re: [HACKERS] PQescapeIdentifier |