11-12-2012 04:03 PM
Hi to all:
We have two switches brocade 5100 and brocade 300 and we are monitoring our SAN via SNMP, with Nagios tool; and we have received alerts constantly with the next message:
CRIT - assuming 2Gbit/s, In: 145.29B/s, Out: 34.29KB/s, no TX buffer credits: 24.60%CRIT, Phy:inSync(6), Opnline(1), Admnline(1).
minutes later other one:
WARN - assuming 2Gbit/s, In: 23.20KB/s, Out: 61.48KB/s, no TX buffer credits: 7.74%WARN, Phy:inSync(6), Opnline(1), Admnline(1);
and for last other one:
OK - assuming 2Gbit/s, In: 5.65KB/s, Out: 5.50KB/s, Phy:inSync(6), Opnline(1), Admnline(1).
The performance is normal; has not been degraded.
Messages appear in: 2 ports connected to HBAs, and an other two ports connected to storages.
So, the question is: If really there is something wrong with our environment?
Regards from Mexico.
11-12-2012 10:24 PM
I've made in the past diverse test with Nagios and other Commercial and Free SNMP Browser , and noted follow.
I get 3 diverse error message!
what is want to say is, if the link/port work fine, you should ignore this message, maybe test the same with another SNMP Browser to see if the alert is continued.
Is my opinion.
11-13-2012 12:41 AM
I am not a Nagios expert and let us assume that the collected values are correct. I would say that the ports where you have 25% of the time "no buffer credits" and showing small throughput like the above I would assume that the device is really slow or is overloaded with simultaneous open requests.
If you say the devices are operating as expected and everybody is happy with the performance you should think to change your thresholds within Nagios. So increase the No TX Buffer Credits threshold.
11-21-2012 08:31 AM
11-24-2012 01:02 AM
We are using Nagios and I've heard check_mk but I didn't know that you can check the brocade switches with check_mk. It looks really nice. The next thing I have to test.