25 June 2012 Karim Elatov

I had recently received a call saying that all of the VMs are grayed out in his inventory when logged into the vCenter and checking the storage view doesn’t show any LUNS. When I logged into the system I saw the following:

missing_vms

When I logged into the host I wanted to make sure we are connected to the SAN appropriately, so first I checked which HBAs are active, and I saw the following:

 # esxcfg-scsidevs -a
vmhba0 megaraid_sas link-n/a unknown.vmhba0 (1:0.0) LSI Logic / Symbios Logic Dell PERC 6/i
vmhba1 qla2xxx link-n/a fc.2000001b320b9001:2100001b320b9001 (8:0.0) QLogic Corp ISP2432-based 4Gb Fibre
vmhba3 ata_piix_ link-n/a ide.vmhba3 (0:31.1) Intel Corporation 631xESB/632_ESB IDE
vmhba32 usb-storage link-n/a usb.vmhba32 () USB
vmhba33 usb-storage link-n/a usb.vmhba33 () USB
vmhba34 ata_piix link-n/a ide.vmhba34 (0:31.)} Intel Corporation 631xESB/632_ESB IDE

Checking out the proc node for the card that is showing link-na and is the only non-local HBA (vmhba1), we saw the following:

 # cat /proc/scsi/qla2xxx/7
QLogic PCI to Pibre Channel Host Adapter for QLE2460:
Firmware uersion 4.04.09 [IP] [Multi-ID] [84XX] , Driuer uersion 8.02.01-X1-vmw43
BIOS version 2.02
FCODE version 2.00
EPI version 2.00
Flash PW version 4.03.01
ISP: ISP2432
Request Queue = 0x1d813000, Response Queue = 0x1d894000
Request Queue count = 4096, Response Queue count = 512
Total number of interrupts = 3909
Device queue depth = 0x20
Number of free request entries = 4096
Number of mailbo_ timeouts = 0
Number of ISP aborts = 0
Number of loop resyncs = 1
Host adapter:loop state = <DEAD>, flags = 0x105a83
Dpc flags = 0x40180c0
MBX flags = 0x0
Link down Timeout = 045
Port down retry = 005
Login retry count = 008
Execution throttle = 2048
ZIO mode = 0_6, ZIO timer = 1
Commands retried with dropped frame(s) = 0
Product ID = 0000 0000 0000 0000

NPIV Supported: yes
Max Virtual Ports = 127

SCSI Device Information:
scsi-qla0-adapter-node=2000001b320b9001:000000:0;
scsi-qla0-adapter-port=2100001b320b9001:000000:0;

FC Target-Port List:

FC Port Information:

And of course checking for the storage devices the host sees, we saw the following:

# esxcfg-scsidevs -c
Deuice UID Deuice Type Console Deuice Size Plugin Display Name
mpx.vmhba32:C0:T0:L0 CD-ROM /dev/sr0 0MB NMP Local USB CD-ROM (mpx.vmhba32:C0:T0:L0)
mpx.vmhba33:C0:T0:L0 Direct-Access /dev/sda 0MB NMP USB Direct-Access (mpx.vmhba33:C0:T0:L0)
mpx.vmhba3:C0:T0:L0 CD-ROM /dev/sr1 0MB NMP Local HL-DT-ST CD-ROM (mpx.vmhba3:C0:T0:L0)
naa.6001e4f0349cda000fa0bf4703f2279a Direct-Access /dev/sdb 69376MB NMP Local DELL Disk (naa.6001e4f0349cda000fa0bf4703f2279a)

Nothing but local storage was seen.The reason why the VMs were showing up grayed out was because the HBA used to connect the SAN was down.

We moved the HBA around and it was discovered that the original HBA went bad and now the stand-by HBA had to be used. After we had moved all the cables around, we saw link up from the esx host for the good HBA:

 # cat /proc/scsi/qla2xxx/7
QLogic PCI to Pibre Channel Host Adapter for QLE2460:
Firmware uersion 4.04.09 [IP] [Multi-ID] [84XX] , Driuer uersion 8.02.01-X1-vmw43
BIOS version 2.02
FCODE version 2.00
EPI version 2.00
Flash PW version 4.03.01
ISP: ISP2432
Request Queue = 0x1d813000, Response Queue = 0x1d894000
Request Queue count = 4096, Response Queue count = 512
Total number of interrupts = 3909
Device queue depth = 0x20
Number of free request entries = 4095
Number of mailbo_ timeouts = 0
Number of ISP aborts = 0
Number of loop resyncs = 1
Host adapter:loop state = <READY> , flags = 0x105ac3
Dpc flags = 0x0
MBX flags = 0x0
Link down Timeout = 045
Port down retry = 005
Login retry count = 008
Execution throttle = 2048
ZIO mode = 0_6, ZIO timer = 1
Commands retried with dropped frame(s) = 0
Product ID = 0000 0000 0000 0000

NPIV Supported: yes
Max Virtual Ports = 127

SCSI Device Information:
scsi-qla0-adapter-node=2000001b320b9001:000000:0;
scsi-qla0-adapter-port=2100001b320b9001:000000:0;

FC Target-Port List:

FC Port Information:

But now we had to modify the zoning on the SAN switch to allow the new WWN of the HBA to have access to the EMC SAN. The customer was using an MDS 9020 Cisco San switch:

san_switch# show hardware

Cisco MDS 9000 FabricWare
Copyright (C) 2002-2005, by Cisco Systems, Inc.
and its suppliers. All rights reserued.
Copyrights to certain works contained herein are owned by
third parties, and used and distributed under license.
Portions of this software are gouerned by the GNU Public License,
which is available at http://www.gnu.org/licenses/gpl.html.

Software
system: 2.1(3)
system compile time: Wed Jul 5 13:21:55 2006

Hardware
san_switch uptime is 381 days 4 hours 14 minute(s) 3 second(s)

Last reset at 32933643 usecs after Thu Nou 13 19:29:26 2008
Reason: ResetByInstaller

------------------------------
Switch hardware ID information
------------------------------

MDS Switch is booted up
Model number is DS-C9020-20K9
H/W uersion is 2.0
Part Number is xxxx xx
Part Reuision is xxx
Serial number is xxxxxxxx
CLEI code is xxxxxx

First we needed to make sure that we see our WWN logged into the switch:

san_switch# show flogi database
-------------------------------------------------------------------
INTERFACE FCID PORT_NAME NODE_NAME
-------------------------------------------------------------------

fc1/3 0x760200 50:03:08:c0:9c:00:80:01 50:03:08:c0:9c:00:80:00
fc1/4 0x760300 21:00:00:e0:8b:93:8d:5a 20:00:00:e0:8b:93:8d:5a
fc1/7 0x760600 21:00:00:1b:32:0b:46:01 20:00:00:1b:32:0b:46:01
fc1/8 0x760700 21:00:00:1b:32:82:54:af 20:00:00:1b:32:82:54:af
fc1/9 0x760800 21:00:00:1b:32:82:f4:b2 20:00:00:1b:32:82:f4:b2
fc1/10 0x760900 21:00:00:1b:32:0b:d8:00 20:00:00:1b:32:0b:d8:00
fc1/11 0x760a00 21:00:00:1b:32:0b:10:01 20:00:00:1b:32:0b:10:01
fc1/12 0x760b00 21:00:00:1b:32:0b:91:00 20:00:00:1b:32:0b:91:00
fc1/13 0x760c00 21:00:00:1b:32:0b:2a:01 20:00:00:1b:32:0b:2a:01
fc1/14 0x760d00 50:06:01:68:41:e0:b9:44 50:06:01:60:c1:e0:b9:44
fc1/15 0x760e00 50:06:01:60:41:e0:b9:44 50:06:01:60:c1:e0:b9:44
fc1/17 0x761000 21:00:00:e0:8b:93:56:37 20:00:00:e0:8b:93:56:37
fc1/19 0x761200 21:00:00:1b:32:0b:49:01 20:00:00:1b:32:0b:49:01
fc1/20 0x761300 21:00:00:1b:32:0b:90:01 20:00:00:1b:32:0b:90:01

Total number of flogi = 14

The last line matches the WWN from the above output of the proc node information from the host:

SCSI Device Information:
scsi-qla0-adapter-node=2000001b320b9001:000000:0;
scsi-qla0-adapter-port=2100001b320b9001:000000:0;

The customer was utilizing fcaliases, zones, and zonesets for his zoning, pretty standard practice on the MDS Series. More information regarding fcaliases, zones, and zonesets can be found in the article entitled “Cisco’s Manual of the switch. So we had to modify the fcalias (basically an alias for a wwn, but has other functions as well) and remove the original pWWN and add the new one:

san_switch# conf t
Enter configuration commands, one per line.

san_switch(config)# fcalias name ESXHOST3_VMHBA2
san_switch(config-fcalias)# no member pwwn 21:00:00:1b:32:0b:8f:01
san_switch(config-fcalias)# member pwwn 21:00:00:1b:32:0b:90:01
san_switch(config-fcalias)#

After modifying the fcalias we double checked that the *zones * (collection of fcaliases that can talk to each other) configuration has been updated:

san_switch# show zones
zoneset name zonesetnew
...
...
Zone name ESXHOST_3_VMHBA2
fcalias name ESXHOST_3_VMHBA2
pwwn 21:00:00:1b:32:0b:90:01
fcalias name CX3-40C_SPA
pwwn 50:06:01:60:41:e0:b9:44
fcalias name CX3-40C_SPB
pwwn 50:06:01:68:41:e0:b9:44

That looked good, now looking at the active zoneset (collection of zone configurations that define access to the whole fabric, to better understand differences between zones and zoneset check out Figure 12-3 Hierarchy of Zone Sets, Zones, and Zone Members from the Manual of the MDS SAN Switch) we saw the following.

san_switch# show zoneset active
zoneset name zonesetnew
...
...
zone name ESXHOST_3_VMHBA2
pwwn 21:00:00:1b:32:0b:8f:01
pwwn 50:06:01:60:41:e0:b9:44
pwwn 50:06:01:68:41:e0:b9:44

It was still showing the old pWWN. We had to commit the changes that we had made by activating the zoneset:

san_switch)# conf t
san_switch(config)# zoneset activate name zonesetnew
Zoneset activation initiated. check zone status
san_switch(config)# end
san_switch#

After that we ran show zoneset active and we didn’t see any asterisks (*) next to any zones, which means that all the zones loaded up just fine. And also after doing a rescan from the ESX host, all the original LUNs showed up on the host and all the grayed out VMs came be into inventory without any issues.


blog comments powered by Disqus