Xymon Mailing List Archive search

Missing events on server's history page?

5 messages in this thread

list Betsy Schwartz · Wed, 22 Apr 2015 09:59:52 -0400 ·
Running xymon 4.3.17 , on linux, wondering if anyone's seen anything like
this?
for example our "edapps2-cf2" just alerted on cpu.

The non-green view shows these most recent event

Wed Apr 22 09:33:25 2015    edapps2-cf2.example.com    cpu
(yellow->green)
Wed Apr 22 09:28:24 2015    edapps2-cf2.example.com    cpu    (red->yellow)
Wed Apr 22 09:18:22 2015    edapps2-cf2.example.com    cpu
 (yellow->red)
Wed Apr 22 09:08:19 2015    edapps2-cf2.example.com    cpu
 (green->yellow)


When we go to the host page then cpu dot,and look at the full history we
don't see  them

All log entries
Date    Status    Duration
Mon Apr 13 17:00:44 2015    8 days 16:40:47      (green)
Mon Apr 13 16:45:41 2015    0:15:03                    (yellow)
Tue Mar 24 19:11:14 2015    19 days 21:34:27    (green)

When we look at Reports/Event log they show up

Wed Apr 22 09:43:45 2015     edapps2-cf2.example.com     http     green
From -> To     yellow
Wed Apr 22 09:33:25 2015     edapps2-cf2.example.com     cpu     yellow
From -> To     green
Wed Apr 22 09:28:24 2015     edapps2-cf2.example.com     cpu     red From
-> To     yellow
Wed Apr 22 09:18:22 2015     edapps2-cf2.example.com     cpu     yellow
From -> To     red
Wed Apr 22 09:08:19 2015     edapps2-cf2.example.com     cpu     green From
-> To     yellow

why aren't they on the host/cpu history page?
thanks for any clues/explanation. We do see other events with shorter
duration on that history page
list Johan Björklund · Wed, 22 Apr 2015 16:44:40 +0200 ·
Hi,

I've saw something like this quite a few times at my previous employer
(This was with an ancient 4.2.x installation). The problem in that case was
that the specific history event file for the test had a negative timed
duration entry (normally duration was -1 second ... ). Looking at the entry
after, calculate and insert a correct duration and you should get your
complete history log for the specific test in the web gui.

In your case I wold check out: $XYMONHISTDIR/edapps2-cf2,example,com.cpu

regards
Johan Björklund

On Wed, Apr 22, 2015 at 3:59 PM, Betsy Schwartz <user-c61747246f66@xymon.invalid>
quoted from Betsy Schwartz
wrote:
Running xymon 4.3.17 , on linux, wondering if anyone's seen anything like
this?
for example our "edapps2-cf2" just alerted on cpu.

The non-green view shows these most recent event

Wed Apr 22 09:33:25 2015    edapps2-cf2.example.com    cpu
(yellow->green)
Wed Apr 22 09:28:24 2015    edapps2-cf2.example.com    cpu
(red->yellow)
Wed Apr 22 09:18:22 2015    edapps2-cf2.example.com    cpu
 (yellow->red)
Wed Apr 22 09:08:19 2015    edapps2-cf2.example.com    cpu
 (green->yellow)


When we go to the host page then cpu dot,and look at the full history we
don't see  them

All log entries
Date    Status    Duration
Mon Apr 13 17:00:44 2015    8 days 16:40:47      (green)
Mon Apr 13 16:45:41 2015    0:15:03                    (yellow)
Tue Mar 24 19:11:14 2015    19 days 21:34:27    (green)

When we look at Reports/Event log they show up

Wed Apr 22 09:43:45 2015     edapps2-cf2.example.com     http     green
From -> To     yellow
Wed Apr 22 09:33:25 2015     edapps2-cf2.example.com     cpu     yellow
From -> To     green
Wed Apr 22 09:28:24 2015     edapps2-cf2.example.com     cpu     red From
-> To     yellow
Wed Apr 22 09:18:22 2015     edapps2-cf2.example.com     cpu     yellow
From -> To     red
Wed Apr 22 09:08:19 2015     edapps2-cf2.example.com     cpu     green
From -> To     yellow

why aren't they on the host/cpu history page?
thanks for any clues/explanation. We do see other events with shorter
duration on that history page

list Betsy Schwartz · Wed, 22 Apr 2015 11:06:19 -0400 ·
Thank you!
those events aren't actually in the file. If they're not in the file how
does the event report see them, which file does that look at?

[root at xymon1 hist]#    pwd
/usr/local/xymon/data/hist
[root at xymon1 hist]# tail edapps2-cf2,exampe,com.cpu
Thu Mar  5 19:19:02 2015 green 1425601142 1395994
Sun Mar 22 00:05:36 2015 red 1426997136 1514
Sun Mar 22 00:30:50 2015 yellow 1426998650 301
Sun Mar 22 00:35:51 2015 green 1426998951 1210
Sun Mar 22 00:56:01 2015 yellow 1427000161 2099
Sun Mar 22 01:31:00 2015 green 1427002260 236015
Tue Mar 24 19:04:35 2015 blue 1427238275 399
Tue Mar 24 19:11:14 2015 green 1427238674 1719267
Mon Apr 13 16:45:41 2015 yellow 1428957941 903
Mon Apr 13 17:00:44 2015 green 1428958844[root at xymon1 hist]#
list Betsy Schwartz · Mon, 4 May 2015 11:41:44 -0400 ·
this is getting to be a huge issue. We have huge gaps in what is showing up
on an individual server's event history
everything shows up in the event log report


On Wed, Apr 22, 2015 at 11:06 AM, Betsy Schwartz <user-c61747246f66@xymon.invalid>
quoted from Betsy Schwartz
wrote:
Thank you!
those events aren't actually in the file. If they're not in the file how
does the event report see them, which file does that look at?

[root at xymon1 hist]#    pwd
/usr/local/xymon/data/hist
[root at xymon1 hist]# tail edapps2-cf2,exampe,com.cpu
Thu Mar  5 19:19:02 2015 green 1425601142 1395994
Sun Mar 22 00:05:36 2015 red 1426997136 1514
Sun Mar 22 00:30:50 2015 yellow 1426998650 301
Sun Mar 22 00:35:51 2015 green 1426998951 1210
Sun Mar 22 00:56:01 2015 yellow 1427000161 2099
Sun Mar 22 01:31:00 2015 green 1427002260 236015
Tue Mar 24 19:04:35 2015 blue 1427238275 399
Tue Mar 24 19:11:14 2015 green 1427238674 1719267
Mon Apr 13 16:45:41 2015 yellow 1428957941 903
Mon Apr 13 17:00:44 2015 green 1428958844[root at xymon1 hist]#

list Japheth Cleaver · Thu, 7 May 2015 00:29:24 -0700 ·
quoted from Betsy Schwartz

On Mon, May 4, 2015 8:41 am, Betsy Schwartz wrote:
this is getting to be a huge issue. We have huge gaps in what is showing
up
on an individual server's event history
everything shows up in the event log report


On Wed, Apr 22, 2015 at 11:06 AM, Betsy Schwartz
<user-c61747246f66@xymon.invalid>
wrote:
Thank you!
those events aren't actually in the file. If they're not in the file how
does the event report see them, which file does that look at?

[root at xymon1 hist]#    pwd
/usr/local/xymon/data/hist
[root at xymon1 hist]# tail edapps2-cf2,exampe,com.cpu
Thu Mar  5 19:19:02 2015 green 1425601142 1395994
Sun Mar 22 00:05:36 2015 red 1426997136 1514
Sun Mar 22 00:30:50 2015 yellow 1426998650 301
Sun Mar 22 00:35:51 2015 green 1426998951 1210
Sun Mar 22 00:56:01 2015 yellow 1427000161 2099
Sun Mar 22 01:31:00 2015 green 1427002260 236015
Tue Mar 24 19:04:35 2015 blue 1427238275 399
Tue Mar 24 19:11:14 2015 green 1427238674 1719267
Mon Apr 13 16:45:41 2015 yellow 1428957941 903
Mon Apr 13 17:00:44 2015 green 1428958844[root at xymon1 hist]#

Hi Betsy,

History lines will be present in three distinct files (assuming you've
enabled the various options for saving events):

In your case, you should see, for each 'cpu' event on this host, entries in:

1) /usr/local/xymon/data/hist/edapps2-cf2,exampe,com.cpu
2) /usr/local/xymon/data/hist/edapps2-cf2.exampe.com
3) /usr/local/xymon/data/hist/allevents


It should look something like this:
[root at rhel6-x86-64 hist]# tail -n 3 f21.build f21,build.conn
==> f21.build <==
conn 1430851017 1430828798 22219 gr re 1
conn 1430851048 1430828798 22250 re gr 2
conn 1430851081 1430828798 22283 gr re 1

==> f21,build.conn <==
Tue May  5 04:38:48 2015 red 1430825928 2870
Tue May  5 05:26:38 2015 green 1430828798 22283
Tue May  5 11:38:01 2015 green 1430851081[root at rhel6-x86-64 hist]#

[root at rhel6-x86-64 hist]# grep f21 allevents | grep conn | tail -3
f21.build conn 1430828802 1430828798 4 re gr 2
f21.build conn 1430851048 1430828798 22250 re gr 2
f21.build conn 1430851081 1430828798 22283 gr re 1


Do you notice any unusual log lines for the relevant tasks? Alternatively,
are you low on disk space on the volume xymond_history is running on? By
default, it's configured to not save full snapshots of status logs when
less than 5% is remaining.


On any restart of xymond, xymond will send a current status report through
xymond_history, which will open up each host+svc event file and validate
the last color matches what's currently set (updating if needed -- if the
color changed while xymond was turned off), which *should* fix a missing
event record.

It might also help to have xymond_history in --debug mode, to see if
there's a potential logic flaw in how it's writing out.


HTH,

-jc