Unsolved

This post is more than 5 years old

10 Posts

1084

January 3rd, 2007 21:00

Help with FC Login to SP

Hello,
I have about one year experience with our CX-300 and MCDATA, and am one of the Storage Engineers. I have successfully zoned and attached many hosts to the CX-300, with no FC login problems. But now I have two recently attached Windows 2003-R2 hosts to the MCDATA switches/SAN fabric, and these two hosts show "Logged in to SP- NO Registered - YES" in the Navisphere Management Utility. We have FLARE 19, Navisphere agents, and Powerpath. I believe the HBA's are logging in to the MCDATA FC switches just fine, but are not logging in to the CX-300 SP's. Powerpath shows no problems. Do I have a host configuration security login issue from the host based Navisphere agent to the SP, or do I have a zoning problem? Thanks, JSB

10 Posts

January 3rd, 2007 21:00

PS. To Clarify:

In "Host Connectivity" in Navisphere Management Utility,

HBA links of two Windows 2003-R2 hosts show:
"Logged in - NO Registered - Yes"

9 Legend

 • 

20.4K Posts

January 4th, 2007 05:00

so at one point they were logged in and registered and now they are not logged in but still registered? At this point you can't see any LUNs, can you ?

10 Posts

January 4th, 2007 07:00

PS. Yes, they are always registered, but not logged in.

10 Posts

January 4th, 2007 07:00

Hello,
LUN's can be seen. I'm not sure if the hosts ever logged into the SP. Powerpath shows active pathways from both hosts and all links to SAN. Any thoughts? What is the problems that can be associated with a registered host, but not logged into an SP?

Thanks,
JSB

9 Legend

 • 

20.4K Posts

January 4th, 2007 07:00

this is strange behavior for a windows box. My HPUX boxes tend to logout while access to LUNs is still available, but have yet to see any of my windows boxes log out while still providing LUN access. Were these hosts registered manually or with navi agent ?

4 Operator

 • 

2.1K Posts

January 4th, 2007 08:00

I have to agree this is very strange. In almost 5 years and 14 CLARiiON arrays I have never seen this happen on anything other than HP-UX (just don't get me started on THOSE boxes!!).

Can you confirm that the WWNs of the HBAs you have successfully logging into the switches are the same as the WWNs of the initiators that show Registered but not Logged In?

10 Posts

January 4th, 2007 09:00

Yes, the WWN's are the same. But I do have an occasional red blinking light on one of the two hosts on both of that hosts HBA's. The green light is on constantly, and the red blinks on occasion on both HBA's. Communication errors?

9 Legend

 • 

20.4K Posts

January 4th, 2007 09:00

Do you use Connectrix Manager to manage your McData's? If yes, can you look in Element Manager and see if those ports have "Link Incident" ? or any other Alerts?

10 Posts

January 4th, 2007 10:00

Some Progress. I un-did the zoning, and re-did the zoning, and Initiators are now logging in to SP's properly. So that problem was doing a repair to zoning records. Now I just have to deal with the blinking orange/red on the one particular hosts HBA's. No link errors according to "Connectrix" basic on the MCDATA's. Maybe when we hard set the speed and port type on the HBA's to match the hard settings of the MCDATA ports, that HBA problem will clear up.

Thanks for all your help so far.
JSB

10 Posts

January 4th, 2007 10:00

Set the host's Mcdata port back to auto-negotiate, and the blinking orange/red still happens on the active cluster node's HBA's. The HBA on the passive cluster node doesn't blink Orange/Red. They are Qlogic HBA's, but the blinking light is not amber, I am sure. It is Orange, or Red.

Thanks,
JSB

9 Legend

 • 

20.4K Posts

January 4th, 2007 10:00

do you see any problems with HBA driver in device manager on that system? Is this host active node in the cluster right now ? Can you set port speed on McData back to auto-negotiate and see what happens?

4 Operator

 • 

2.1K Posts

January 4th, 2007 10:00

Even with Connectrix Manager Basic on the switches you can still look at the Link Incident Log to see if there are synchronization errors.

And I would like to get clarification of the blinking light on the HBA. If it is an amber light blinking at regular intervals, this just indicates that you have a valid FC link. I only deal with Emulex and qLogic HBAs, but everything I've seen will have one green light on to indicate power and one flashing amber LED when you get a good FC link to the HBA.

10 Posts

January 4th, 2007 10:00

We have "Connectrix Basic" on the MCDATA's, nothing more. No port errors though. Just blinking Orange/Red on the one hosts HBA ports. Link Utilization is just 1%. Other host in two host cluster, (currently "passive" mode in the MS Cluster), doesn't blink Orange/Red, but also doesn't have any I/O through it. I just finished setting all MCDATA ports to both hosts to "F Port", "2 GB", but won't have a chance to set the HBA's from Auto-Negotiate to "F Port", "2 GB" until down time allows. Seems strange though if that it.

Any other idea's?

JSB

10 Posts

January 4th, 2007 10:00

Okay, failed over the virtual cluster node from the active node to the passive node, and now the opposite node, which is now the active cluster node, shows the blinking orange/red on the HBA's. They are Qlogic HBA's. So now I'm wondering if the HBA's Orange/Red blinking lamp is just a "write" blink. The previously active cluster node, that is now passive, is not blinking orange/red anymore. Any thoughts?

Thanks,
JSB

9 Legend

 • 

20.4K Posts

January 4th, 2007 12:00

are these QLA 2340/2342 ?

0 events found

No Events found!

Top