Re: BUG #5946: Long exclusive lock taken by vacuum (not full)

Dimitri Fontaine <dimitri@2ndquadrant.fr>

From: Dimitri Fontaine <dimitri@2ndQuadrant.fr>
To: Greg Stark <gsstark@mit.edu>
Cc: Tom Lane <tgl@sss.pgh.pa.us>, Alvaro Herrera <alvherre@commandprompt.com>, Christopher Browne <cbbrowne@gmail.com>, Maxim Boguk <maxim.boguk@gmail.com>, pgsql-bugs <pgsql-bugs@postgresql.org>
Date: 2011-05-02T19:05:25Z
Lists: pgsql-bugs
Greg Stark <gsstark@mit.edu> writes:
> So is it ok for inserting one row to cause my table to grow by 90GB?
> Or should there be some maximum size increment at which it stops
> growing? What should that maximum be? What if I'm on a big raid system
> where that size doesn't even add a block to every stripe element?

I'd think that capping the idea to the segment size makes sense.

Also, what about having a background process (bgwriter or autovacuum
come to mind) doing the work, rather than the backend that happens to be
inserting the row?

It could send a message, and continue creating a newer 8kb block if the
background process has not yet extended the storage.  Also, to be safe I
guess we could arrange to have the new segment be created way before
reaching the very end of the relation (so that adding 8kb does not need
to create a new segment, so as to avoid a race condition with the
background process doing so itself).

Regards,
-- 
Dimitri Fontaine
http://2ndQuadrant.fr     PostgreSQL : Expertise, Formation et Support