pg.ddx.io pgsql-admin@postgresql.org mailing list archive
help / color / mirror / Atom feedDeadlock on "select ... for update"?
3+ messages / 2 participants
[nested] [flat]
* Deadlock on "select ... for update"?
@ 2011-11-29 18:15 Craig James <craig_james@emolecules.com>
0 siblings, 1 reply; 3+ messages in thread
From: Craig James @ 2011-11-29 18:15 UTC (permalink / raw)
To: pgsql-admin
Several times recently one of our databases has gotten stuck with the
following situation:
postgres=# select datname, procpid, usename, current_query from
pg_stat_activity where current_query != '<IDLE>';
datname | procpid | usename |
current_query
------------+---------+----------+--------------------------------------------------------------------------------------------------------
emolecules | 13503 | customerdb | select tableid from hitlist_table_pool
where hitlistid <= 0 for update
emolecules | 32082 | customerdb | select tableid from hitlist_table_pool
where hitlistid <= 0 for update
emolecules | 17974 | customerdb | select tableid from hitlist_table_pool
where hitlistid <= 0 for update
emolecules | 31299 | customerdb | select tableid from hitlist_table_pool
where hitlistid = 0 limit 1 for update
emolecules | 30247 | customerdb | select tableid from hitlist_table_pool
where hitlistid = 0 limit 1 for update
postgres | 1705 | postgres | select datname, procpid, usename,
current_query from pg_stat_activity where current_query != '<IDLE>';
emolecules | 28866 | customerdb | <IDLE> in transaction
emolecules | 21394 | customerdb | select tableid from hitlist_table_pool
where hitlistid = 0 limit 1 for update
emolecules | 22237 | customerdb | select tableid from hitlist_table_pool
where hitlistid = 0 limit 1 for update
(9 rows)
It's obvious that they're all waiting ... but for what? The "<IDLE>"
process looks like the culprit, but how do I figure out what it's doing?
The next time this happens, what queries can I run to help diagnose what's
going on?
This is PG 8.4.4 on Ubuntu 10.
Thanks,
Craig
^ permalink raw reply [nested|flat] 3+ messages in thread
* Re: Deadlock on "select ... for update"?
@ 2011-11-29 18:36 Scott Marlowe <scott.marlowe@gmail.com>
parent: Craig James <craig_james@emolecules.com>
0 siblings, 1 reply; 3+ messages in thread
From: Scott Marlowe @ 2011-11-29 18:36 UTC (permalink / raw)
To: Craig James <craig_james@emolecules.com>; +Cc: pgsql-admin
On Tue, Nov 29, 2011 at 11:15 AM, Craig James
<craig_james@emolecules.com> wrote:
> Several times recently one of our databases has gotten stuck with the
> following situation:
>
> postgres=# select datname, procpid, usename, current_query from
> pg_stat_activity where current_query != '<IDLE>';
> datname | procpid | usename |
> current_query
> ------------+---------+----------+--------------------------------------------------------------------------------------------------------
> emolecules | 13503 | customerdb | select tableid from hitlist_table_pool
> where hitlistid <= 0 for update
> emolecules | 32082 | customerdb | select tableid from hitlist_table_pool
> where hitlistid <= 0 for update
> emolecules | 17974 | customerdb | select tableid from hitlist_table_pool
> where hitlistid <= 0 for update
> emolecules | 31299 | customerdb | select tableid from hitlist_table_pool
> where hitlistid = 0 limit 1 for update
> emolecules | 30247 | customerdb | select tableid from hitlist_table_pool
> where hitlistid = 0 limit 1 for update
> postgres | 1705 | postgres | select datname, procpid, usename,
> current_query from pg_stat_activity where current_query != '<IDLE>';
> emolecules | 28866 | customerdb | <IDLE> in transaction
> emolecules | 21394 | customerdb | select tableid from hitlist_table_pool
> where hitlistid = 0 limit 1 for update
> emolecules | 22237 | customerdb | select tableid from hitlist_table_pool
> where hitlistid = 0 limit 1 for update
> (9 rows)
>
> It's obvious that they're all waiting ... but for what? The "<IDLE>"
> process looks like the culprit, but how do I figure out what it's doing?
> The next time this happens, what queries can I run to help diagnose what's
> going on?
> This is PG 8.4.4 on Ubuntu 10.
Does this help?
http://wiki.postgresql.org/wiki/Lock_Monitoring
^ permalink raw reply [nested|flat] 3+ messages in thread
* Re: Deadlock on "select ... for update"?
@ 2011-11-30 18:08 Craig James <craig_james@emolecules.com>
parent: Scott Marlowe <scott.marlowe@gmail.com>
0 siblings, 0 replies; 3+ messages in thread
From: Craig James @ 2011-11-30 18:08 UTC (permalink / raw)
To: Scott Marlowe <scott.marlowe@gmail.com>; +Cc: pgsql-admin
On 11/29/11 10:36 AM, Scott Marlowe wrote:
> On Tue, Nov 29, 2011 at 11:15 AM, Craig James
> <craig_james@emolecules.com> wrote:
>> Several times recently one of our databases has gotten stuck with the
>> following situation:
>>
>> postgres=# select datname, procpid, usename, current_query from
>> pg_stat_activity where current_query != '<IDLE>';
>> datname | procpid | usename |
>> current_query
>> ------------+---------+----------+--------------------------------------------------------------------------------------------------------
>> emolecules | 13503 | customerdb | select tableid from hitlist_table_pool
>> where hitlistid<= 0 for update
>> emolecules | 32082 | customerdb | select tableid from hitlist_table_pool
>> where hitlistid<= 0 for update
>> emolecules | 17974 | customerdb | select tableid from hitlist_table_pool
>> where hitlistid<= 0 for update
>> emolecules | 31299 | customerdb | select tableid from hitlist_table_pool
>> where hitlistid = 0 limit 1 for update
>> emolecules | 30247 | customerdb | select tableid from hitlist_table_pool
>> where hitlistid = 0 limit 1 for update
>> postgres | 1705 | postgres | select datname, procpid, usename,
>> current_query from pg_stat_activity where current_query != '<IDLE>';
>> emolecules | 28866 | customerdb |<IDLE> in transaction
>> emolecules | 21394 | customerdb | select tableid from hitlist_table_pool
>> where hitlistid = 0 limit 1 for update
>> emolecules | 22237 | customerdb | select tableid from hitlist_table_pool
>> where hitlistid = 0 limit 1 for update
>> (9 rows)
>>
>> It's obvious that they're all waiting ... but for what? The "<IDLE>"
>> process looks like the culprit, but how do I figure out what it's doing?
>> The next time this happens, what queries can I run to help diagnose what's
>> going on?
>> This is PG 8.4.4 on Ubuntu 10.
> Does this help?
>
> http://wiki.postgresql.org/wiki/Lock_Monitoring
Yes, thanks! That's exactly what I needed.
Craig
^ permalink raw reply [nested|flat] 3+ messages in thread
end of thread, other threads:[~2011-11-30 18:08 UTC | newest]
Thread overview: 3+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2011-11-29 18:15 Deadlock on "select ... for update"? Craig James <craig_james@emolecules.com>
2011-11-29 18:36 ` Scott Marlowe <scott.marlowe@gmail.com>
2011-11-30 18:08 ` Craig James <craig_james@emolecules.com>
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox