Tom Lane escribió:
> Alvaro Herrera <alvherre(at)commandprompt(dot)com> writes:
> > Gregory Stark escribi:
> >> A crazy idea I just had -- what if you roll this into the deadlock check? So
> >> after waiting on the lock for 1s it wakes up, finds that the holder it's
> >> waiting on is an autovacuum process and cancels it instead of finding no
> >> deadlock.
> > Another crazy idea is to have some sort of "blacklist" of tables in
> > shared memory. Any autovacuum process would skip those tables.
> The deadlock check idea sounds promising to me, not least because it
> avoids adding any cycles in performance-critical paths. I'm not certain
> how easy it'd be to fold the idea into the checker though. That
> logic is pretty complicated :-( and I'm not sure that it makes a
> consistent effort to visit every possible blocker.
The idea sounds interesting, but I am not at all sure how to fit it in
the deadlock code.
I am totally uninclined to mess with this stuff. I am barely aware of
what exactly is it doing; I don't have the slightest idea how to modify
it to cancel autovacs. Furthermore it sounds very much like a layering
violation (what is deadlock.c doing with autovac processes anyway).
Alvaro Herrera http://www.advogato.org/person/alvherre
"Pensar que el espectro que vemos es ilusorio no lo despoja de espanto,
sólo le suma el nuevo terror de la locura" (Perelandra, CSLewis)
In response to
pgsql-hackers by date
|Next:||From: Simon Riggs||Date: 2007-10-04 18:40:46|
|Subject: Re: First steps with 8.3 and autovacuum launcher|
|Previous:||From: Tom Lane||Date: 2007-10-04 18:22:40|
|Subject: Re: Not *quite* there on ecpg fixes |