Craig A. James a écrit :
> I sent the attached email ony to Jan by mistake, here it is, I hope Jan
> or somebody has an idea what's going on.
> 
> A quick summary:
>  1. Shut of Apache
>  2. Restart Postgres and Slony
>  3. lock set (id = 1, origin = 1);
>     wait for event (origin = 1, confirmed = 2);
>     move set (id = 1, old origin = 1, new origin = 2);
> 
> When I do this, Slony starts a SELECT that never finishes ("never" means
> I waited 10 minutes, 35 minutes, and 10 minutes in three different
> tests, with no sign of progress).
> 
> Per Jan's request, I've attached the output of two SQL queries:
> 
>    select relname, granted, pid, mode from pg_locks L, pg_class C
>        where C.oid = L.relation and locktype = 'relation';
> 
>    select * from pg_stat_activity;
> 
> In addition, I've attached the output of vmstat, a few snapshots taken
> over a 30-minute peroid.  It's looks to me like postgres is doing some
> sort of huge join or sort, based on the changing amounts of CPU and I/O
> during the 30 minutes I monitored it.  No other processes were active;
> this was entirely due to the backend Postgres process being used by Slony.
> 
>> From these, you can see that a Slon process is doing a "SELECT
>> moveSet(...)".  ps(1) confirms that this process is the one using all
>> the CPU time shown by vmstat.
> 
> I sure could use some help on this one, it's become a critical roadblock
> to our operation.
> 
> Thanks,
> Craig
> 

Hi, Not sure I got the full story as this thread began with vacuum and then
came to move set...

But, concerning move set hanging, I remember some experienced the problem. So
had I. I then stoppped the shell having initiated the move set, and restartzed
it (I have to test if set is locked...).

Second time, move set was ok.

I surely don't know if that could be of any help.

Cheers,

-- 
Stéphane SCHILDKNECHT
Président de PostgreSQLFr
Tél. 09 53 69 97 12
http://www.postgresqlfr.org