SSA cable Connections for HACMP training system

[wpdavids]

I am setting up a test system 2 node ver5.4 HACMP cluster running AIX 5.3 and need some help with cabling a SSA 7133-D40. I have 2 - 44p-170 workstations. Each workstation has a 6230 SSA adapter with 4 ports. I have 8 - 9GB drives and would like information on a basic SSA cable design. If I only have 8 drives I think I will need fillers.

Please help I am looking to educate Unix administrators at a not for profit organization.

 

[DukeSSD]

Filters ???

The SSA redbook probably has what you are after:

http://www.redbooks.ibm.com/abstracts/sg245750.html

Stick all the drives in the front of the drawer and connect one hosts A loop to ports 1 and 8 and the other hosts A loop to 4 and 5 - unless you want to get fancy with morroring, etc. and split the drawer and use two loops on each host.

 

[wpdavids]

Yes.. I was asking if I need disk drive fillers? for the empty slots in the storage array. I only have 8 disks for my 7133-D40.
I was able to cable the SSA loops like DukeSSD explained and the drives can be seen with "lsdev -Cc disk", however, one drive did not become available. Also when I try to create a VG I get a failure and the VG can not be created. After trying to create the VG almost all of the SSA disks go into defined state.

I'm really having trouble understanding what I am doing wrong.

 

[DukeSSD]

Form memory - diag, task selection, ssa service aids.
Check your loops are ok and check the disk status.

What errors did you see?

 

[p5wizard]

For SSA cabling, see this:

http://publib.boulder.ibm.com/infoc...?topic=/com.ibm.itame.doc/am61_install362.htm

specifically:
- An SSA-based disk array, such as: IBM® 7133 Model T40 storage tower or an IBM 7133 D40 rack mounted enclosure
- Three SSA connection cables. Two (one per AIX system) are cabled to the disk array and one is cabled between the two AIX systems

first cable: system1 SSA adapter A1 connector to "1" connector of SSA drawer
second cable: "8" connector of SSA drawer to system2 SSA adapter A2 connector
third cable: system2 SSA adapter A1 connector back to system1 SSA adapter A2 connector.
(provided your eight disks are in slots 1 to 8.)

With 3 more cables you can do the same setup with 2 strings of 4 disks (connector "1" and "4" for first string and "5" and "8" for second string) and then you also use the B1/B2 connectors of the SSA adapters.

The empty disk modules or fillers are not needed to close the SSA loops as you only use connectors "1" and "8" (connectors "4" and "5" are automatically connected to each other inside the SSA drawer if you do not plug in a "live" SSA cable). But the fillers may still be needed for correct cooling air flow.



HTH,

p5wizard

 

[wpdavids]

This is the second configuration I tried.

first cable: system1 SSA adapter A1 connector to "1" connector of SSA drawer
second cable: "8" connector of SSA drawer to system2 SSA adapter A2 connector
third cable: system2 SSA adapter A1 connector back to system1 SSA adapter A2 connector.
(provided your eight disks are in slots 1 to 8.)

I can then see 7 ssa disks available and 1 as defined. I was only able to create a volume group with two ssa disks. When trying to create one with all 7 disks I got an error. 0516-1182 open failure on hdisk and 0516-862 unable to create VG.

 

[DukeSSD]

Do you know the history of the drives?
If they were previously in a raid(5) set you will need to format them ecause the bytes per sector are different for raid and non-raid disks.
Typically a defined disk is not responding and has failed.
Try running "diag > advanced diags > system verification" select each disk in turn and then F7 to test, take the option to certify the media in each case to check the disk is OK.
Only format them as a last resort because this will reset the bad block reallocation and lead to an early life failure when they are put back into use.
I think you can see the formatted bytes per sector, probably in the lasttr -El output or the disk which might give a clue, I think non-raid are 512 and raid are 522.
Is the disk that is not responding showing a light on the control panel of the 7133?
I think you can probably see the disk and link status in the "diag > task selection > ssa service aids" tool.

 

[wpdavids]

Thanks for all your help.

First off all the SSA executable files did not have executable permissions. This gave me problems managing the SSA disks from smitty devices and caused problems for the system to configure the drives running "cfgmgr" on reboot.

I did format the drives... and when doing this noticed errors from some of the drives stating they are part of a raid array.

For the SSA drives previously configured for a raid array - I went into smitty ssaraid and removed the raid settings on the drives.

Rebooted the system and all the drives came up available and I could configure them in a VG.

I used p5wizard's SSA loop recommendations and I was able to configure HACMP with an Enhanced Concurrent Capable VG, did a import of the VG, synchronized the cluster, did a resource group move, and Everything worked.

Thanks again for you tips!!!

 

[p5wizard]

Did you act on this too?

The empty disk modules or fillers are not needed to close the SSA loops ... [red]But the fillers may still be needed for correct cooling air flow.[/red]




HTH,

p5wizard

 

[wpdavids]

Yes..I did get 8 empty disk modules and filled the back section of the SSA draw. I don't expect to add more drives anytime soon (this is only for testing HACMP) so the loops should be ok. However, anything else you would like to mention would be appreciated.

For right now I am thrilled that my project is moving along. We only have a production HACMP cluster and having one for testing and training administrators will really help us out.
The only thing missing was having a disk array to configure a highly available VG.

Thanks again for the tips because they guided me to finding the root cause of the problems and lead me to the solution.