On Jun 10, 2008, at 1:44 PM, anastrophe wrote:
this strikes me as a bug....but what do i know. zenoss 2.2.0, centos
5.1. since i'm still learning the software, i figured 'hey, why
don't i monitor sendmail on the mta farm we have?'. great idea.
create 'sendmail' on the /processes page. remodel one of the mta
servers, it picks up a boatload of sendmail processes. lovely! oops
- it seems it interprets each instance uniquely - so
sendmail: qdir.0/m5A8CCYj020927 live.net.: user open
is picked up - then when that instance dies off (because the mail's
been delivered), i start getting alerts. oh noes!
so, smart guy goes to the /processes page, and deletes sendmail
there, and remodels the mta server. no more sendmail processes on
the device OS page. that should do it, right? nope. i keep getting
alerts. wtf? i poke and prod, i remodel, i restart zenoss, but can't
seem to stop the alerts. finally, this morning, it dawns on me.
recreate 'sendmail' under /processes. remodel the mta server. the
device shows a bunch of sendmail processes again. okay. delete all
those processes on the OS tab, then go to the /processes page, and
*then* delete sendmail there. the alerts stop.
this *seems* like a bug to me. continuing to get alerts for
processes being down, when they don't show up on the OS process list
for the device. it's kind of like something stays in place in the
code, but hidden from view.
am i right, or is that desired behavior?
You're right that this is a bug. I just attempted to reproduce it, and
was easily able to do so. Even deleting the process directly from the
OS tab didn't cause the monitoring to stop for me. Only restarting
zenprocess after making the changes would work. I created the ticket
linked below to document the problem.
Ticket #3408: zenprocess doesn't get notified of process additions/
deletions
http://dev.zenoss.org/trac/ticket/3408
_______________________________________________
zenoss-users mailing list
[email protected]
http://lists.zenoss.org/mailman/listinfo/zenoss-users
