[mrtg] Discontinuity at midnight

[Mathew Marulla]

Steve -

We do have a number of F5s, but they are at different datacenters, not configured in a cluster.  We are seeing the same behavior at each.

I am looking at the statistics on the actual F5s today to see if they show the same pattern.  That would rule out any issues with SNMP & MRTG.

- Matt

On Dec 18, 2012, at 7:00 PM, Steve Shipway <s.shipway at auckland.ac.nz> wrote:

> That is really weird.  We have an F5 too, but don’t experience this sort of pattern.
>  
> It can’t be a counter reset at midnight, because that would only affect the first sample after that point.  The graphs look almost like two separate things glued together, rather than a single event at midnight.
>  
> My only guess would be some sort of scheduled task causing a failover.  Do you have several F5 devices in a cluster?  If so, it could be that at midnight (for some reason) one or more groups fail over to the other cluster member.  This would result in fewer connections via the one you’re monitoring.  Maybe check the logs to see if and when any cluster group failovers happen, or changes between active/active and active/passive on groups.
>  
> I’d be interested to hear what the eventual resolution for this is.
>  
> Steve
>  
> Steve Shipway
> ITS Unix Services Design Lead
> University of Auckland, New Zealand
> Floor 1, 58 Symonds Street, Auckland
> Phone: +64 (0)9 3737599 ext 86487
> DDI: +64 (0)9 924 6487
> Mobile: +64 (0)21 753 189
> Email: s.shipway at auckland.ac.nz
> P Please consider the environment before printing this e-mail : 打印本邮件，将减少一棵树存活的机会
>  
> From: mrtg-bounces+s.shipway=auckland.ac.nz at lists.oetiker.ch [mailto:mrtg-bounces+s.shipway=auckland.ac.nz at lists.oetiker.ch] On Behalf OfMathew Marulla
> Sent: Wednesday, 19 December 2012 1:16 a.m.
> To: mrtg at lists.oetiker.ch
> Subject: [mrtg] Discontinuity at midnight
>  
> Is it normal to see values jump suddenly at midnight?  Is something getting in SNMP or MRTG getting reset?  I am seeing this in both normal and gauge type graphs.  Don't believe there is anything getting reset on the server (an F5 Big-IP load balancer, in this case).
>  
> - Matt
>  
> <image001.png>
>  
> <image002.png>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.oetiker.ch/pipermail/mrtg/attachments/20121219/9d92ab14/attachment.htm