[Zope-DB] Reviving an old thread :: DCO2 and memleaks...
Geir Bækholt
Geir Bækholt <geirh@funcom.com>
Thu, 9 Jan 2003 12:42:08 +0100
Reviving an old thread...
on or about, Thursday, May 02, 2002, we have reason to believe that
Dario Lopez-K=E4sten wrote something along the lines of :
> Hello!
> We have DCO2 on Linux and we are experiencing two problems:
> For certain large queries, zope keeps a connection to the oracle server
> after the query has finished, which over time makes the database server u=
se
> up all it's resources, requiring it to be restarted.
We are seeing the exact same problem, with the latest released DCO2. We
are currently leaking about one connection a minute, requiring us to
restart our Zope every 4 hours or so..
> related to this, this query also seems to make zope/python leak memory.
> usage goes from 23 MB ram to about 200-300 MBs (normal operating mem
> footprint is 25-35 MB).
Our is about 5-700 MB when we kill it. Obviously this is becoming a
rather serious issue..
> Restarting the zope app, makes both the DB server happy, as well as memory
> usage go down to acceptable levels.
I am experiencing the same behaviour.
> Any hint on what might be causeing this=3F Workarounds=3F Known bugs=3F
> I assume that this is not the intended behaviour of DCO2/Zope, even is the
> queries are inane (we know some of the are, and we're working on fixing
> them...)
I am running mostly very mundane queries , not returning many rows,
even, but we do have a fair amount of traffic... (well.. squid eats most
of it)
Did anyone ever find a workaround or fix for this =3F
--=20
Geir B=E6kholt geir@funcom.com
Tools/HCI-developer
Tools/Billing - Product Operations
Funcom Oslo