05-18-2012 12:16 PM
I have problem with disappearing paths between servers and eva 4400. The paths come back after 1-2min or after multipath scan. There is nothing in logs.. It's happens on different physical server, few linux distributions. In most cases Eva losts paths in one fabric but it happend twice that she lost all paths in both fabrics. I looking for some way to do deeper debuging.
The are few ( 6-7) switches in two fabrics:
- 71.2 Brocade 300 it's HP 8/24.
- 72.3 Brocade 5480 it's Brocade 8Gb SAN Switch for HP BladeSystem c-Class
fos most 6.2, 6.3, 6.4 also - i will all upgrade to 7.0.1a in 1 week. There are ca 30 zones in each fabrics
I'd like ask about monitoring tools, i'd like to exclude problems with switches. What tools do you prefer. I think about buy:
Advanced Performance Monitoring
but they cost a little bit, are they good for low level debugging and can help to solve this problem ?
thanks for any help
05-19-2012 01:36 AM
sure they can help. I haven't used them though.
Which logs have you checked?
One thing you could do is to monitor all the port statistics on the ports in the SAN and look for errors.
See output of CLI command portstatsshow.
HP support has a way of making a comparison between the output of two supportshows (so that you see which port error counters are increasing, possibly indicating where the problem might be).
Perhaps DCFM / BNA has some nice features that doesn't cost extra money that can help with this?
I would look in the 'errshow' of all the switches. Maybe there is a switch acting out.
Monitor load and error of ISL links, maybe some gets over-loaded.
I have seen a bad HBA cause problems like this.
About the FabOS versions - are any of them non-latest? For example are you running 6.3.0* ?
See if the problem happens at around a certain time each time.
If possible, gather supportshows/supportsaves right after the problem occured (and look in output of fabriclog -s).