Unsolved

This post is more than 5 years old

19 Posts

1583

July 1st, 2008 10:00

Upgrade from PowerPath 4.0 to 5.1 - Please Share Experience or Tips

Hello All,

Our CLARiiON environment is connected to 29 hosts where 18 are Windows-based hosts (majority are Windows 2003 and all boot from SAN) and 11 ESX 2.5 hosts.
Majority of the Windows-based hosts are connected via QLogic PCI-X HBA (Firmware 1.47) and PowerPath version 4.1.1.
For CLARiiON FLARE Upgrade to the latest update, we're now must upgrade all the PowerPath and QLogic software first.
I have the documentation for PowerPath 5.1 downloaded from Powerlink.
Our first priority is the stability and uptime of the servers.

Can anyone please share their experience or tips-tricks when doing this upgrade, so that we can avoid any problems?
Any critical points that we must consider?
Can a customer with only basic CLARiiON training do this job?
Please share your experience and thoughts.

Many Thanks in Advance.

/JP

11 Legend

 • 

20.4K Posts

 • 

87.4K Points

July 1st, 2008 11:00

Start with E-Lab navigator, put in all the components and see what's supported, what needs to be upgraded, target versions for PowerPath, drivers, firmware. I always start with host side upgrades and do Clariion flare upgrades last. If customer has only basic knowledge ..i would get EMC involved as i am sure data availability is number 1 concern :)


Home > Support > Interoperability and Product Lifecycle Information > E-Lab Interoperability Navigator

6 Operator

 • 

1.5K Posts

July 1st, 2008 13:00

Please take a look at the CLARiiON Software Upgrade steps on Powerlink at the following location -

Home > Support > Product and Diagnostic Tools > CLARiiON Tools > CLARiiON Software Update Tools > CLARiiON Software Update (Standard)

Also - run a Grab report on the hosts and get the HEAT report generated - you can access HEAT now on Powerlink - or else request your EMC team to do the same.

Grab Utilities for all platforms are available on Powerlink at -

Home > Support > Product and Diagnostic Tools > Grab Utilities

and HEAT (Host Environment Analysis Tool) is at -

Home > Support > Product and Diagnostic Tools > Environment Analysis Tools

And of course, follow Dynamox's suggestion and if required get EMC team engaged.

All the very best ..
Sandip

24 Posts

July 2nd, 2008 04:00

while upgrading Powerpath 4.x to 5.1, the Powerpath framework changes. So you need to remove Powerpath completely from the host and then install Powerpath 5.1. Please refer to EMC Knowledgebase article emc90664 to get more information on the procedure how to remove old versions of Powerpath and install new version.

24 Posts

July 2nd, 2008 04:00

sorry .. forgot to mention ... HBA drivers and firmware and MS hotfix 943545 (if the HBA driver page suggest) should be installed prior to installing Powerpath. The procedure should be MS Hotfix, HBA firmware, HBA driver and then Powerpath.

As dynamox suggested before, at any point, you come across any issues, please contact Support.

19 Posts

July 3rd, 2008 12:00

Hello Amit and Everyone,
Thanks for responding with such helpful information.

I get the following procedure from a friend, could you please give comment or verify whether these are in correct sequence:

1. Install the latest QLogic storport driver
2. Install the Microsoft hotfix
3. Upgrade Power Path 4.4.1 to current (uninstall 4.4.1 and install 5.2)
4. Download the latest QLogic BIOS
5. Using the registry editor change the DisktimeoutValue registry key found at HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\Disk from 0x28 to 0x3c

Please kindly advice.
Thank You.

/JP

24 Posts

July 3rd, 2008 20:00

Hi Juang,

As I mentioned earlier, Powerpath framework is different from 4.6.x onwards from versions prior to that. It now works on Microsoft MPIO. So it's always recommended to uninstall any previous versions of Powerpath and then install Powerpath v5.1. The procedure should be:

1) If cluster, work on passive nodes first
2) Reduce the server to one path by leaving just one HBA enabled and the rest all disabled in Device Manager.
3) Uninstall PP .. reboot
4) Install Microsoft Hotfix .. reboot
5) Install EMC approved Qlogic storport driver and firmware (this will fix the registry) these can be downloaded from http://driverdownloads.qlogic.com/QLogicDriverDownloads_UI/Oem_EMC.aspx .. reboot
6) Install PP
7) Enable all the paths
8) Failover the cluster
9) follow steps 2-8 again on the other node.

19 Posts

July 4th, 2008 07:00

Hi Amit,
Thanks for the info.

Is the step No.5 in my previous mail still necessary or can be skipped?

5. Using the registry editor change the DisktimeoutValue registry key found at HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\Disk from 0x28 to 0x3c

Please kindly advice.
Thank You.

/JP

6 Operator

 • 

5.7K Posts

July 4th, 2008 07:00

5) Install EMC approved Qlogic storport driver and firmware (this will fix the registry)


It seems that the install will fix the registry setting....

19 Posts

July 4th, 2008 08:00

RRR, thanks for responding.

Amit, RRR and everyone,
Just one last question from me in this particular thread, regarding steps described by Amit:
1) If cluster, work on passive nodes first
2) Reduce the server to one path by leaving just one HBA enabled and the rest all disabled in Device Manager.
3) Uninstall PP .. reboot
4) Install Microsoft Hotfix .. reboot
etc

Pardon my limited knowledge, but as mentioned on my initial post, all of our servers boot from SAN, so if we do step 2 and step 3 above, the server will lost connection to the SAN and have no access to its own Operating System (Windows-based) and cannot proceed to step 4 etc, am I thinking correctly?

Please kindly advice.
Thanks.

/JP

6 Operator

 • 

5.7K Posts

July 4th, 2008 08:00

Good thinking !
you are completely right. You must make sure the host keeps it's connection working and by uninstalling ppath, you need to make sure only 1 path is working. Correct !

11 Legend

 • 

20.4K Posts

 • 

87.4K Points

July 4th, 2008 15:00

this adds another caveat ..if the server is zoned to SPA/SPB per HBA ..then blocking one HBA will not reduce storage access to one path ..you will still have 2 paths because the other HBA is zoned to both SPs ..hence you still have two paths there. Are all your systems going to be upgraded at the same time ? I am just asking because of what i am about to propose ? I could be missing something so would like to hear feedback from others. What if Juang trespasses all LUNs to SPA, blocks port(s) on the switch for SPB, at this point systems can only get to storage through SPA but we still have each HBA connected to SPA (so still 2 paths), at this point go ahead and block one HBA and perform PP upgrade. What ya'll think ?

24 Posts

July 7th, 2008 01:00

RRR, thanks for responding.

Amit, RRR and everyone,
Just one last question from me in this particular
thread, regarding steps described by Amit:
1) If cluster, work on passive nodes first
2) Reduce the server to one path by leaving just one
HBA enabled and the rest all disabled in Device
Manager.
3) Uninstall PP .. reboot
4) Install Microsoft Hotfix .. reboot
etc

Pardon my limited knowledge, but as mentioned on my
initial post, all of our servers boot from SAN, so if
we do step 2 and step 3 above, the server will lost
connection to the SAN and have no access to its own
Operating System (Windows-based) and cannot proceed
to step 4 etc, am I thinking correctly?

Please kindly advice.
Thanks.

/JP


JP,
You can disbale all paths except one either before or after uninstalling PP, but don't initialize the disk through Disk Management because OS will try to access the LUN from both HBAs/Paths trying to write Signature and may leave your disk unusable.

6 Operator

 • 

5.7K Posts

July 7th, 2008 01:00

The point is that iif you upgrade drivers or anything for that matter on 1 HBA, that HBA will go down and you loose connectivity. So if you do both HBA's at the same time, the server will get a BSOD (if it's Windows) and possibly won't boot. To make sure the server still can boot from SAN, you need 1 HBA functional, so upgrade 1 HBA and leave 1 working as it is and then reverse: do the other HBA and keep the upgraded one to boot from.
Furthermore: without PPath, you should bring down 1 HBA in the first place.

I'm sorry if my english is a bit lala, but I've had a tough weekend ;)

24 Posts

July 7th, 2008 01:00

Sorry for delayed response. It was weekend here in India ...Hope you had a good weekend ... Answer to your query dynamox, the HBA is zoned to both SP A and SP B, the server will have paths through both SPs. But only one SP owns LUN at a given point of time, so there should be just one path at one given time. So Zoning should not be a issue here. You can just either disbale one HBA from Device Manager or just remove the fibre cable.

11 Legend

 • 

20.4K Posts

 • 

87.4K Points

July 7th, 2008 05:00

Sorry for delayed response. It was weekend here in
India ...Hope you had a good weekend ... Answer to
your query dynamox, the HBA is zoned to both SP A and
SP B, the server will have paths through both SPs.
But only one SP owns LUN at a given point of time, so
there should be just one path at one given time.


as far as PowerPath is concerned ..if the HBA is zoned to SPA and SPB there are two paths regardless of which SP is currently the owner of the LUN. So if you block one HBA ..you still have 2 paths. Here is an output from powermt display on a windows box where each HBA is zoned to SPA/SPB:


Pseudo name=harddisk2
CLARiiON ID=APM000XXXXXXXXX [DEV]
Logical device ID=60060160E7040E00118B85399215DA11 [LUN 260]
state=alive; policy=CLAROpt; priority=0; queued-IOs=0
Owner: default=SP A, current=SP A
==============================================================================
---------------- Host ---------------   - Stor -   -- I/O Path -  -- Stats ---
### HW Path                 I/O Paths    Interf.   Mode    State  Q-IOs Errors
==============================================================================
   0 port0\path0\tgt0\lun1     c0t0d1    SP A1     active  alive      0      0
   0 port0\path0\tgt1\lun1     c0t1d1    SP B1     active  alive      0      0
   1 port1\path0\tgt0\lun1     c1t0d1    SP A3     active  alive      0      0
   1 port1\path0\tgt1\lun1     c1t1d1    SP B3     active  alive      0      0
No Events found!

Top