02-22-2012 07:30 AM
I am a newbie regarding san switch. apparently we are having this warning error on our Brocade 300 24 port with Fabric OS 6.4.1a version.
does anyone experienced this warning??
02-22-2012 08:07 AM
Those sort of events usually report a credit loss or a slow drain device situation. What's connected to those ports?
02-22-2012 08:21 AM
Then, it seems that that Bladecenter switch is behaving as a slow drain device (or maybe one of the hosts connected to it). Which firmware versioes does it have? and the Brocade 300?
Check if there are errors in any of the Brocade 5470 (you can attach the output of porterrshow) to see if there is any port reporting errors.
On the other hand, upgrading the Bladecenter Firmware can prevent this situation from occuring. Also setting the speed manually in the 4570 ports help.
02-22-2012 08:31 AM
The Brocade 300 has Fabric firmware 6.4.1a and Brocade 5470 (Bladeswitch) has the same.
the output for bladeswitch for porterrshow is attached. On port 16 and 17 are the ISL connection to Brocade 300 (port 6 and 7).
if you change the speed to 8Gb on the port, would it not affect the HBA that are running 4Gb fixed?? We have fixed alle internally to run 4Gb Hba's.
02-22-2012 08:44 AM
ports 4 and 7 are reporting disc_c3 errors as well as CRC_g_eof. These CRC errors report physical issue (altough some times they can report an HBA driver issue). If you block these two ports, probably the latency events will disappear.
Try clearing the port stats by running statsclear command and after a while get porterrshow again to see which ports errors are actively increasing.
I recommend you to inspect the hba-cable-sfp in ports 4 and 7 and also check for newer HBA drivers, and FOS releases and upgrade if possible.
02-22-2012 08:48 AM
forgot to mention that on the February the 7th we had a total UPS power supply failure on all san switches, which we had to startup manually.
The issues on the bottelnecks started the February 10. It might have something to do with that?
02-22-2012 11:54 AM
I would suggest that you run the command portdisable and portenable on your ISL ports to the Bladecenter.
This reset will give you all buffers back on the ISL ports in case a buffer lost.
Do this in a low traffic time because you will loose some frames and this will cause IO errors on the servers. After that run the command statsclear on your san switch and the BC switch. Then check if you have increasing error counters.
Which speed have your BC servers and are they all connected to old slow storage or tape drives?