PowerPath 5.2 (build 146) Host Initiators Logged in = No

Question

Hi,

Recently we’ve noticed that 5 of our HP blade servers (windows 2003 x64 sp2 / VMWare ESX 4 update 1) are keep logging out (can be for 5 minutes or more) and automatically logging in for the same time.

We haven’t done any changes to the system or to OS.

We tried to restart the brocade switch but unfortunately without a success.

I've looked at the logs (Event Viewer, SP Collect) and couldn’t find any problems.

Please let me know if you had a similar problem or thoughts for the cause of it.

Many thanks,

Eran

Eran2 · Answer

No, the drive is still there, it is normally loosing 2-3 path only

dynamox · Answer

can you isolate it to particular switch fabric only ?

dynamox · Answer

does it happen for all paths at the same time ? So the host pretty much loses access to its disk ?

Eran2 · Answer

When you go to the server conectivies check (from the Navispher): SP port B-0, B-3, B2 (logged in = no) and than it recover it self and from the PowerPath - SP A3 = dead

dynamox · Answer

are B0, B2, B3 and A3 connected to the same switch ? Going through the same patch panel ..something similar between them ?

Eran2 · Answer

They are all connected directly vi FC to the same Brocade switch. I've uploaded 2 screen shots as well (hoping it will make it easier to understand)

Eran2 · Answer

Sorry my mistake, the picture is correct. Just to make it clear after 5 minutes or more it will connect itself automatically and will stay on for 5 minutes or more and will disconnect itself again.

dynamox · Answer

you mentioned that B3 was having this issue, but according to your snapshot your host is logged-in to B3 just fine.

dynamox · Answer

how are your blades connected, do you have switches internal to the enclosure ?

Eran2 · Answer

Thanks for pointing me to the article emc197345,

Unfortunately this is not the case, we haven't done any driver upgrade and as I wrote earlier the path are connecting itself automatically and disconnect itself automatically in circle with no specific reason.

Eran2 · Answer

The two Brocade switches (Brocade 4/12 SAN Switch) are both internal interconnect cards in the blade chassis (HP BLc7000). The blade server models are ProLiant BL460c G1, We've upgraded the firmware of the switch the latest one but unfortunately it is still the same

dynamox · Answer

so you see any fabric events when the host is logging-in and logging out ?

kelleg · Answer

I'd call HP and have them review the switch log - if you see the initiators logged out on the clariion, it's a good bet that the connection was dropped - mostly likely at the switch - are you using hard (port) zoning or soft (WWN) zoning?

glen

Eran2 · Answer

Hi All,

Thanks for helping me,

I've checked the logs of the Brocade switch and found an error when I run the errdump:

2010/07/02-04:47:36, [PORT-1003], 100, CHASSIS, WARNING, xxx, Port 14 Faulted because of many Link Failures

2010/07/06-08:43:30, [PORT-1003], 101, CHASSIS, WARNING, xxx, Port 6 Faulted because of many Link Failures

I only found an article on the HP website: http://h200003.www2.hp.com/bizsupport/TechSupport/Document.jsp?lang=en&cc=us&taskId=110&prodSeriesId=439556&prodTypeId=12169&objectID=c01460227 but our HBAs are mezzanine cards in a blade and not PCI.

sadik1 · Answer

Looking at all the above conversation it is definitely an issue outside the CLARiiON. I would suggest contacting HP support for a faster resolution. I would also consider replacing the FC cables connecting the SP ports on SPA and SPB as well. Using different ports (on the switch) for SP connection would also be a good considerable step to isolate physical Switch port issues.

CLARiiON

PowerPath 5.2 (build 146) Host Initiators Logged in = No

Was this post helpful?