Xymon Mailing List Archive search

problem with setting time window for alerts

4 messages in this thread

list Elizabeth Jones · Mon, 18 Mar 2013 10:08:47 -0500 ·
I'm apparently doing something extremely wrong trying to set a time window for alerts for a specific server. This server burns up cpu every night and we don't want to see alarms for it, so I tried doing this to block those alarms in my hosts.cfg file -

IP   servername # conn NKTIME=*:0800:2200

I thought this was working but all of a sudden this weekend we got a ton of alarms from it again during the night.
list Ryan Novosielski · Mon, 18 Mar 2013 11:34:22 -0400 ·
Isn't NKTIME deprecated? I don't use it anyway, but do use this feature. I believe the documentation mentions this.
quoted from Elizabeth Jones


From: Jones, Elizabeth [mailto:user-f0dffc744f19@xymon.invalid]
Sent: Monday, March 18, 2013 11:08 AM
To: xymon at xymon.com <xymon at xymon.com>
Subject: [Xymon] problem with setting time window for alerts


I’m apparently doing something extremely wrong trying to set a time window for alerts for a specific server. This server burns up cpu every night and we don’t want to see alarms for it, so I tried doing this to block those alarms in my hosts.cfg file –

IP   servername # conn NKTIME=*:0800:2200

I thought this was working but all of a sudden this weekend we got a ton of alarms from it again during the night.
list Steve Holmes · Mon, 18 Mar 2013 11:48:38 -0400 ·
quoted from Ryan Novosielski
On Mon, Mar 18, 2013 at 11:08 AM, Jones, Elizabeth <user-f0dffc744f19@xymon.invalid> wrote:
I’m apparently doing something extremely wrong trying to set a time window
for alerts for a specific server. This server burns up cpu every night and
we don’t want to see alarms for it, so I tried doing this to block those
alarms in my hosts.cfg file –

IP   servername # conn NKTIME=*:0800:2200

I thought this was working but all of a sudden this weekend we got a ton
of alarms from it again during the night.

NKTIME won't do what you want. It only selects when something shows up on a
critical systems page (and is deprecated anyway).

Instead ...

If you want to inhibit the high cpu showing up on the web pages (and
therefore inhibit the alert) use the TIME= parameter on a high LOAD value
for that host in analysis.cfg. E.G.:

LOAD 1200 1500 HOST=db.foo.com TIME=2200:0800

If you just want to inhibit the alerts (but let the red icon still show up
on the web pages) you can do it in the alerts.cfg file.

HOST=db.foo.com SERVICE=cpu
  MAIL recipient TIME=0800:2200

Which means that alerts will only be sent for cpu on db.foo.com to
recipient between 8AM and 10PM.

Steve Holmes
Purdue
list Mike Burger · Mon, 18 Mar 2013 11:50:18 -0400 (EDT) ·
-- 
Mike Burger
http://www.bubbanfriends.org

"It's always suicide-mission this, save-the-planet that. No one ever just
stops by to say 'hi' anymore." --Colonel Jack O'Neill, SG1
quoted from Steve Holmes

I'm apparently doing something extremely wrong trying to set a time window
for alerts for a specific server. This server burns up cpu every night and
we don't want to see alarms for it, so I tried doing this to block those
alarms in my hosts.cfg file -

IP   servername # conn NKTIME=*:0800:2200

I thought this was working but all of a sudden this weekend we got a ton
of alarms from it again during the night.
Hi, Elizabeth.

Try this, instead:

DOWNTIME=cpu:*:hhmm:hhmm:"Reason"

I use this for the backups on my servers, as the CPU spikes during those
operations. Instead of a yellow or red, during those timeframes, it shows
blue and does not send out alerts.