Have you verified with a sniffer that the pings are not reaching the
routers? In our case it was that the routers (mostly older 4000's)
were configured to place ICMP response priority at the bottom of
the list, and when the CPU utilization reached a certain threshold
the routers simply stopped responding to pings. This was a very
transient condition, and usually within a few minutes the routers
would start responding again. Our fix was to upgrade to 4700's
with sufficient memory, and to place ICMP at a higher priority...
BELFORTI Gabriel DICEI wrote:
>
> Hi,
> Occasionally some pings from netmon deamon do not reach the nodes.
> There is no specific pattern about the nodes and time it occurs. It is all
> at random. Most of the nodes are Cisco routers (less than 60 routers) spread
> out around the world.
> When you see the trapd.log file there are many one-minute 'Interface
> down/up' events from different nodes. The polling interval is one minute.
> The retry has been set to 5 times at 10 sec interval.
> In addition, if you ping the node from the map at the same time this
> problem occurs (you see it red on the map), Netview 'realizes' about the
> mistake and changes the status to normal inmediately.
> Does anybody experience the same problem?
> IBM tech support checked our RS6000 and it's OK. Besides we have
> another monitoring platform using the same Catalyst Switch that it's
> connected our RS6000 and it works OK.
> Our RS6000 P240 has 768MB RAM, 9.1Gb disk and dual processor.
>
> Thanks and regards...
>
> Gabriel Belforti
> baibeg@techint.net
> TE: +54-11-4318-2384
> FAX: +54-11-4318-2690
> VIDEO (ISDN): +54-11-4591-2000 (x2)
> http://www.techintgroup.com
|