I saw latency bottleneck events loggin in RASlog, but bottleneck is disabled from config settings. see blow:
Line 37271: 2014/12/28-11:44:58, [AN-1010], 5792, FID 128, WARNING, sanswitch81, Severe latency bottleneck detected at slot 0 port 0.Line 37275: 2014/12/28-12:15:20, [AN-1010], 5794, FID 128, WARNING, sanswitch81, Severe latency bottleneck detected at slot 0 port 0.Line 59774: bottleneck.BECreditLossFaultingBlade:0Line 59775: bottleneck.BECreditLossLRThresh:2Line 60436: [Bottleneck Configuration]Line 60437: bottleneck.enabled:0
It seems bottleneck alerts will be generated to logs no matter what bottleneck setting is. What extra function does the bottleneckmon enables?
The bottleneckmon process when enabled allows you to perform actions based on defined policies via fabricwatch/MAPS. The AN-xxxx messages are simply status messages and are not tied to the fact if bottleneckmon is enabled or not. (I know, a bit confusing.)
But of course that doesn't mean that they don't mean anything. Better see it this way: The bottlenecks logged without enabled bottleneckmon are so severe (thus the "severe" in the message) that the switch feels you should be alerted anyway. Better try to solve them.
Thanks. But what is the definition of "severe" ? Bottleneckmon has its own threshold settings to give alerts, are they any different from the threshold being used with bottleneckmon disabled?
If these AN-1010 messages show up it means that FOS has detected a device that will cause frame flow issues sooner or later which may be causing back-pressure onto ISL's and their respective VC's if not treated. The fact that you can configure your own thresholds does not mean you can circumvent physics.
In FOS 7.4 you can use MAPS SDDQ (Slow Drain device Quarantine) which means that if you can configure a policy that will put a slow-drain device onto a low priority virtual channel and hence not impact other (medium or high QoS) traffic. Very nifty feature.
This is very clear explanation. Thank you.