CVS 1.1.4 seems to have addressed this, great work Matt.

Ted

would /etc/sysconfig/i18n having
SUPPORTED="en_US.iso885915:en_US:en"

as opposed to

SUPPORTED="en_US.UTF-8:en_US:en"

make a difference ?

Regards,
Ted

I setup my dev box to mirror my other box minus customcat and extra
categories/viewdisplay, also the main box has ICMP polls at 30 seconds and one
custom tcp service monitor at 30 seconds as well. The dev box is using all stock
300 second polls. The dev box db is not a mirror of the other box but fairly
close. It appears to have no excessive load issues when deleting a node. Also
noticed that rescans seem much faster on the dev box as well. I started
collecting vital's on the dev box but am fairly certain it does not see these
loads as nothing made me check it Since I now have something decent to
compare with will continue with trying to find what the major trigger is. Will
gather more info as I make the dev box mirror the one I see the loads with.
The dev box is a p700, old ide with 512mb, the other box is P4 2.4, scsi, with 1
gig.

Two notes on delete nodes:

1) We need to build a delete interface and delete service method

2) Check out the performance of SELECT and IN statements

Suspect the fact that opennms is deleting services that never existed on the
node are adding extra load. I make this assumption based upon the events I see
being generated after a noe is deleted. I have an additional 7 custom pollers in
my case using tcp, which probable magnifies the problem.

Regards,
Ted