Chris Macneill via observium wrote on 24/05/2019 15:46:

Thanks for the quick response, but Observium CE GUI is displaying result of SNMPv2-MIB::sysUpTime.0 even when HOST-RESOURCES-MIB::hrSystemUptime.0 is available AND is larger value.

Currently I have three devices displaying the "wrong" System Uptime, 2 x Windows Server 2012R2 and a Tycon Voltage monitor.

Since the last CE release, we have made some improvements regarding uptime,
but main logic (about max uptime) used over the years.

Probably something other wrong in snmp requests..
Attach debug for poller (for problem devices):

./poller.php -r -d -m os -h <device>

The Windows Servers definitely return both values as noted in my original email, but the SNMPv2-MIB::sysUpTime.0 value is displayed.  The Windows Servers are Virtual Machines running under Hyper-V.

I was doing some maintenance on these servers earlier this evening and rebooted them and now the two OIDs return uptime values that are less then 1 minute apart. I guess in a day or so we'll see what happens.

The Tycon monitor has a poor SNMP implementation and only returns SNMPv2-MIB::sysUpTime.0 Observium periodically detects a reboot. I'll try monitoring the OID via a script to try and determine whether the device is providing spurious data or some issue in Observium. I'm not overly concern about this device as it'll get replaced soon, so I may just turn off the alerts. It's more of an irritation factor than a big problem.

Which OS detected on Tycon monitor and around which uptime value reboot triggered for this device?
We have the ability to exclude some "black hole" uptimes from "rebooted" trigger.

It seems to me that for the Windows Servers this may be a bug, so I guess I should report it through the bug report channel? I'll monitor for a few days and then report or not if it continues to be a problem.


Regards 

Chris Macneill 
Web: www.cmit.nz

----- Original Message -----
From: "Mike Stupalov" <mike@observium.org>
To: "observium" <observium@observium.org>, "observium" <observium@observium.org>
Cc: "Chris Macneill" <cmacneill@cmit.nz>
Sent: Friday, 24 May, 2019 22:01:38
Subject: Re: [Observium] System Uptime (sysUpTime) OID incorrect for some systems

sysUpTime is mean:
SNMPv2-MIB::sysUpTime.0
it's complete same as DISMAN-EVENT-MIB::sysUpTimeInstance

$ snmptranslate DISMAN-EVENT-MIB::sysUpTimeInstance -On
.1.3.6.1.2.1.1.3.0
$ snmptranslate SNMPv2-MIB::sysUpTime.0 -On
.1.3.6.1.2.1.1.3.0

Observium always uses maximum time of one of this (which avialable on 
device):
  1. SNMPv2-MIB::sysUpTime.0
  2. HOST-RESOURCES-MIB::hrSystemUptime.0
  3. SNMP-FRAMEWORK-MIB::snmpEngineTime.0
  4. unix-agent or os-specific Oid (from definitions).

In your case will used
hrSystemUptime.0 = Timeticks: (13560578) 1 day, 13:40:05.78


Reboot trigger will happen if in next poll runtime this value less than 
previous more than 300 seconds.


Chris Macneill via observium wrote on 24/05/2019 10:17:

I'm observing a problem with some systems being reported as having 
rebooted at odd times, which are completely unrelated to reality. The 
GUI claims the system uptime is derived from the OID sysUpTime, 
however I can't find this OID in any MIB, maybe an internal label in 
Observium?

In actuality Observium seems to be using the 1st OID below, but as can 
be seen from the 2nd OID the two seem totally unrelated. In the case 
of this particular system, the 2nd OID value is when it was rebooted, 
the 1st one seems to have no correlation to the reboot time.

DISMAN-EVENT-MIB::sysUpTimeInstance = Timeticks: (376806) 1:02:48.06
HOST-RESOURCES-MIB::hrSystemUptime.0 = Timeticks: (13560578) 1 day, 
13:40:05.78

Is there anyway to change the OID on a per device basis that is used 
to display and monitor System Uptime?

Observium CE         18.9.9420 (5th September 2018)
OS         Linux 4.9.0-8-amd64 [amd64] (Debian 9 (stretch))
Apache         2.4.25 (Debian)
PHP         7.0.33-0+deb9u3 (OPcache: ENABLED)
Python         2.7.13
MySQL         10.1.37-MariaDB-0+deb9u1 (extension: mysqli 5.0.12-dev)
SNMP         NET-SNMP 5.7.3
RRDtool         1.6.0
Fping         3.15 (IPv4 and IPv6)

Regards

Chris Macneill
Web: www.cmit.nz



_______________________________________________
observium mailing list
observium@observium.org
http://postman.memetic.org/cgi-bin/mailman/listinfo/observium