Navisphere Agent will not register host with CX

Navisphere Agent will not register host with CLARiiON array

Primus Self-Service

EMC Knowledgebase"Navisphere Agent will not register host with CLARiiON array"
ID: emc125015Usage: 70Date Created: 12/30/2005Last Modified: 06/13/2006STATUS: ApprovedAudience: Customer  Knowledgebase Solution 



Question: How to register a host in Navisphere Manager with Navisphere Agent Environment: Product: CLARiiON CX-Series Environment: OS: Solaris Environment: OS: HP-UX Environment: OS: IBM AIX Environment: OS: Microsoft Windows Environment: OS: Linux Environment: Host adapter: Emulex Environment: Host adapter: QLogic Problem: HBAs log in to Connectivity Status but do not register with the host name. Problem: Host does not register with CLARiiON array. Problem: Restarting the Navisphere Agent does not register the host. See emc95542 ("How to restart Navisphere Agent"). Root Cause: Hostregistration to a CLARiiON array is done in band, that is, through theFibre Channel connection.  In order for a host to register successfullya LUNZ device must be successfully presented to the host for every paththat is zoned to the array.  A LUNZ is a logical device that allowshost software, that is, the Navisphere agent to pass commands to thearray.  LUNZ takes the place of HLU 0 until real physical storage isallocated. The LUNZ devices are created based on the arraycommpathsetting in Navisphere. See solution emc65060 ("What is LUN_Z or LUNz"). Fix: All Host Platforms

1. Verify the agent.config file has the following entries. See emc101128 for host  location. 
2. • device auto auto
   • user system@<IP_SPA>
   • user system@<IP_SPB> 
3. Verify that the Navisphere agent release running on the hostis within two revisions of the FLARE code release running on the array.
4. In order for the host to be a managed host on the CLARiiON array itmust be in the same subnet as the CLARiiON array. See solution emc69147 for a list of TCP service ports required for Navisphere Agent and SP agent Communication. 

Sun Solaris

Emulex HBAs

Verify the following conditions have been met:

• Persistent Bindings have been set up in the lpfc.conf file.  See solution emc53265 ("How to set up Persistent Bindings with lputil"). 
• • fcp-bind-WWPN="5006016023e4578a:lpfc0t16",
  •                             "5006016823e4578a:lpfc0t17"; 
• The sd.conf file has been set up to reflect the lpfc.conf file. 
• • LUNs for lpfc0 target 16
  • name="sd" class="lpfc" target=16 lun=0;
  • name="sd" class="lpfc" target=16 lun=1;
  • etc....
  • LUNs for lpfc0 target 17
  • name="sd" class="lpfc" target=17 lun=0;
  • name="sd" class="lpfc" target=17 lun=1;
  • etc....

What is important is making sure that you have a LUN 0 set up forthe targets established in the lpfc.conf file. If the above is true,the format command should show some devices listed as"Drive Type Unknown." These are the LUNZ devices.  At this pointrestarting the agent should register the host.  If you do not see theLUNZ devices, check the zoning or the host may need a reboot -- -r or both. 

Sun-branded QLogic Cards  (Leadville):

Apply the following procedure:

Add the required system variable in the /etc/system file.

1. set fcp:ssfcp_enable_auto_configuration=1
2. reboot -- -r

If this variable is already in place, follow these steps:

1. Check Navisphere Connectivity Status to make sure all paths are Fibre Logged In. 
2. On the Solaris host run cfgadm.
3. On the Solaris host run devfsadm.
4. Check format to make sure that the LUNZ devices are now available.
5. Restart the Navisphere Agent on the host. 

Linux

Emulex or QLogic HBAs:

The following must be verified on the host:

• The /etc/hostfile has only one entry: The Host Name and IP Address
• The HostIDFile.txt shows the correct IP Address followed by a date stamp
• A LUNZ device is seen on every path to the array in the /proc/scsi/scsi file. If it is not, check zoning and reboot the host.

Note:  If you use QLogic HBAs, make sure that failover is disabled for those HBAs. Make sure that the ql2xfailover and ConfigReguired parameters are set to 0 in the the /etc/modprobe.conf file as shown in the following entry:

options qla2xxx ql2xfailover=0 ConfigRequired=0

Try applying the following procedure:

1. Delete the HostIDFile.txt and restart the Navisphere Host Agent.
2. Reboot the host.
3. Restart Management Server on the array. (In a browser SP_IP/setup => Restart Mgmt Server.)

If above fails, follow these steps:

1. Manually register the host in Navisphere with the Group Edit button in Connectivity Status.
2. Add host and LUNs to the Storage Group.
3. Restart the Navisphere agent or reboot the host.

When manually registering, the host will show "(ManuallyRegistered)" next to its name in Navisphere.  After restarting theagent on the host the "(Manually Registered)" entry will disappear andthe host will now be managed.

Also see solution emc115979 ("Registering Linux host using Navisphere Host Agent") and solution emc60655 ("How to manually register an HBA in Red Hat Linux server"). 

HP-UX

HP  HBAs

Note:  Even if zoning is correct, unlike other Operating Systems you will not see the HBA WWN logging in to Connectivity Status.

Apply the following procedure:

1. Make sure that DNS is enabled in the /etc/nsswitch.conf file!
2. Run ioscan -fnC disk (Scan the SCSI bus for any SCSI devices)
   
   Note: Make sure that you see LUNZ devices.  If you do not, verify the zoning. 
3. Run insf -e.  (Build the special device files for the LUNZ devices.)
4. Restart the Navisphere Host agent.  (Push the information to the array through the newly built LUNZ devices.)

Note:  Even when the host registers it will stillshow Fibre Logged In: No.  This is normal for HP-UX.  It will turn toYes when I/O is going to the allocated devices.  See solution emc99131 ("HP-UX HBA does not appear as 'logged in' from Navisphere").

        If using only PVLinks, verify the following:

• The agent.config file has an "OptionsSupported AutoTrespass (Uncommented)" entry.
• The Initiator Type in Navisphere Manager should be set to "HP auto Trespass."

         If using PowerPath, verify the following:

• The agent.config file has an "#OptionsSupported AutoTrespass" entry.
• The Initiator Type in Navisphere Manager should be set to "HP no-auto Trespass." 

IBM AIX

IBM Branded Emulex HBAs

Caution! Verify the CLARiiON ODM definitions are installed.  They are available on EMC's FTP Site.

1. Run cfgmgr (You may not see the host logged in until you scan the bus with cfgmgr.)
2. Type lsdev -Cc disk | grep LUNZ
3. If there are no LUNZ devices, check the zoning.
4. Restart the Navisphere Host agent to register the host.
5. rmdev -dl hdiskn where n is the hdisk # for the LUNZ devices
6. In Navisphere use the Tools => Failover Setup Wizard to disable arraycommpath for this host.

Note: LUNZ devices on AIX can fill the error reportwith SCSI_DISK ERROR 2's.  Since arraycommpath is the setting that isused to present the LUNZ, it is recommended that you disable thissetting on the chance the HLU 0 is removed from the Storage Group inthe future.  See solution emc111635 ("How to remove LUNZ devices"). 

Microsoft Windows

Emulex or QLogic HBAs

1. Check Disk Management and make sure you see Unknown/Unreadable Devices.
2. • If not, check the zoning.
   • If the zoning appears to be perfect, the host may need to be rebooted. 
3. Restart the Navisphere Host agent (located in Windows Services).

Note: If using Emulex cards with the SCSI FC Port Driver, follow these steps:

1. Open a cmd window and CD into the C:/Program Files/HBAnyware directory.
2. Type: elxcfg.exe --emc
3. Highlight the adapter instance and select Configuration => Set
4. Select the appropriate setting. (This will require a reboot to take affect.)