[rrd-users] Re: Problems with Counters after reboot of device
cliftonr at lava.net
Wed Aug 23 21:37:46 MEST 2000
On Wed, Aug 23, 2000 at 10:39:46AM +0200, Tobias Oetiker wrote:
> Today you sent me mail regarding [rrd-users] Problems with Counters after...:
> *> Hi,
> *> in a lot of my monthly reports i use Counters. The devices i query via SNMP
> *> have to reboot sometimes. After the reboot the counter values restart at
> *> zero.
> *> Through this the later counter values are less then the earlier values.
> *> Rrdttool is convinced that there was a counter wrap. So i get extremly high
> *> values for the the days with the reboots and the values of the other days
> *> are vanishing in the graphs.
> *> Who has an idea what can i do?
You could mimic the logic Cricket uses, which includes also querying
all SNMP devices for "uptime" and expecting counters to be cleared if
it finds the uptime is less than the time since the last sample. I'm
not sure what it does in this case - just stores one unknown value?
At any rate, that's the bare skeleton of one solution.
> the the MIN and MAX properties on all Data sources ... or use
> DERIVE instead of counter and set the MIN property of the DS to 0
> this will ignore all cases where the counter is smaller as the
> previous reading
There has been heated discussion about this on the Cricket list
recently, as to which of these options would be better, mainly for
dealing with the case where counters are being cleared manually by an
operator, or are otherwise being cleared without indication of a device
Do you have an opinion, with regards to oids such as InOctets which can
not meaningfully be negative?
Clifton Royston -- LavaNet Systems Architect -- cliftonr at lava.net
The named which can be named is not the Eternal named.
Unsubscribe mailto:rrd-users-request at list.ee.ethz.ch?subject=unsubscribe
Help mailto:rrd-users-request at list.ee.ethz.ch?subject=help
More information about the rrd-users