Xymon Mailing List Archive search

Lost procs

3 messages in this thread

list Christopher Seip · Sat, 22 Jun 2019 02:21:13 +0000 ·
For one of my Xymon clients running xymon-client, the procs column has disappeared.

My narrative of how this started & what we did lacks some clarity. I think procs was visible last week. If it was there I'd bet a paycheck the status was clear. One day last week it went purple.

So we began troubleshooting & fiddling with it, puzzled by the purple status. I was able to drop the procs test for that host, and it vanished from the web UI.

Then I defined a PROC test for this host in analysis.cfg. We do all our client configuration on the Xymon server and avoid the local configs of Xymon clients. This PROC test, in fact the PROC column for this group (of which only this one host is locally monitored), isn't showing up. Hm.

We've been poking at this puzzle long enough to ask for some help, please. Any thoughts?

--Chris-


CLIENT: RHEL 6.0, Xymon-client 4.3.28-1

SERVER: Ubuntu 16.04.6, Xymon server 4.3.25-1

hosts.cfg excerpt:
    group-compress <I>blah blah </I>
    group-except files|msgs|ports
    16.93.15.15        hostname.domain.com # ssh http://hostname.domain.com/accelerator/images/logo.gif
    include more-hosts.cfg

analysis.cfg excerpt:
HOST=hostname.domain.com
    PROC /sge        3   5  red
list Michael T Pins · Sat, 22 Jun 2019 06:42:28 -0600 (MDT) ·
Seip, Christopher writes:
For one of my Xymon clients running xymon-client, the procs column has disa=
ppeared.
quoted from Christopher Seip
My narrative of how this started & what we did lacks some clarity. I think =
procs was visible last week. If it was there I'd bet a paycheck the status =
was clear.
Which means you didn't have any process checks for it defined.  Which is
normal.
One day last week it went purple.
Which means the client stopped reporting the process data (generally the
output of a `ps` command.
So we began troubleshooting & fiddling with it, puzzled by the purple statu=
s. I was able to drop the procs test for that host, and it vanished from th=
e web UI.
Then I defined a PROC test for this host in analysis.cfg. We do all our cli=
ent configuration on the Xymon server and avoid the local configs of Xymon =
clients. This PROC test, in fact the PROC column for this group (of which o=
nly this one host is locally monitored), isn't showing up. Hm.
Which means that client still isn't reporting the output of `ps`.
quoted from Christopher Seip
We've been poking at this puzzle long enough to ask for some help, please. =
Any thoughts?
Look at the clientlog column for that client.  Do you see anything under
"[ps]"?

Is there anything interesting in the xymon logs on the client (usually, but
not always, in ~xymon/client/logs).
list Christopher Seip · Mon, 24 Jun 2019 16:30:53 +0000 ·
quoted from Michael T Pins
Look at the clientlog column for that client.  Do you see anything under
"[ps]"?
Should have mentioned, yes. The "[ps]" section of clientlog is fully populated.
quoted from Michael T Pins
Is there anything interesting in the xymon logs on the client (usually, but
not always, in ~xymon/client/logs).
I hadn't checked the client logs, good idea! Nothing interesting there, just logfetch barking when the syslog/messages file gets rotated.

Thanks for your thoughts!

--Chris--