Tim,

how do we confirm that the CIM agent is on the master console?

We get an error that says " Error while executing discover LUNS command" , I'm thinking the CIM agent is the issue.

Thoughts?

I think it may be active perl issue. Should we co-locate the active perl with the SVC SRA scripts?

how do we confirm that the CIM agent is on the master console?
Try to telnet to the masterconsole ip on port 5989, if the port respons, the CIM agent is active. From cluster version 4.3.1 the CIM agent is embedded into the cluster firmware and you are able to use the cluster ip address with port 5989 as cim agent. The master console CIMON can be started and stopped from the start menu at the SSPC or master console server.

When using the cluster embedded CIMON you can stop the master console CIMON to reduce the performance impact.

We get an error that says " Error while executing discover LUNS
command"
This is probably an issue with your SRA. This can be because perl isn't running correctly. Try opening up a cmd and see if the "perl" commando is recognised. If this is working in a random directory perl is working correctly. Otherwise this might be a CIM agent issue. Are you able to retrieve the array list? An array for SRM is a cluster, if you are able to see the cluster that is being managed by the cluster or master console cim agent, the cim connection works. Try to reinstall the script package that is being used to be able to communicate between SRM and SVC.

We don't get any reciprocity between the Master consoles however and the
IBM doc is incomplete so we are not 100% sure what we should see.
I know that on the EL arrays the field fill in.
you only need to give the cluster ip address, the cimon address (can be cluster or embedded cimon, depends on version), and the cimon login credentials. When working with the master console cimon agent, you need to use the superuser credentials, when working with the cluster embedded cimon, you are able to use the superuser credentials but I would advice to create another user to be able to troubleshoot the auditlog. I have created seperate users for TPC and SRM.

If I look at my audit log when doing a failover test scenario i receive the following lines:

SRM Test screnario SVC Commands
101:090320170108:<Unknown>::127.0.0.1:srm:0:2:svctask mkvdisk -iogrp 0
-mdiskgrp 1 -size 53687091200 -unit b -vtype striped -cache none
102:090320170109:<Unknown>::127.0.0.1:srm:0:0:svctask mkfcmap -source 1
-target 2 -copyrate 0
103:090320170109:<Unknown>::127.0.0.1:srm:0::svctask prestartfcmap 0
104:090320170333:<Unknown>::127.0.0.1:srm:0::svctask startfcmap 0
105:090320170333:<Unknown>::127.0.0.1:srm:0::svctask mkvdiskhostmap -host
0 -scsi 2 -force 2
106:090320170333:<Unknown>::127.0.0.1:srm:0::svctask mkvdiskhostmap -host
1 -scsi 2 -force 2

SRM Test scenario cleanup
109:090320171140:<Unknown>::127.0.0.1:srm:0::svctask stopfcmap -force 0
110:090320171143:<Unknown>::127.0.0.1:srm:0::svctask rmfcmap -force 0
111:090320171143:<Unknown>::127.0.0.1:srm:0::svctask rmvdisk -force 2

you are able to see that I have created a CIM user named SRM, and that this user is used to be able to execute commands for SRM functions. To troubleshoot the CIM agent connection, activate the more detailed log using CLI or GUI and retrieve the log by using the CLI.

I think it may be active perl issue. Should we co-locate the active perl with the SVC SRA scripts?
Again try opening up a cmd and check if the "perl" commando works. Sometimes rebooting after installing perl can help. If that does not work, check if the perl is in your PATH variable. Here is some help http://technet.microsoft.com/en-us/library/aa997841(EXCHG.65).aspx

tferris,

Any update for the IBM SVC & SRM issue you discovered??

I'm about to log a VMware call, the discoverlun process fails at 23%, i've tryed everything and the logs don't explain that much (discoverLuns exited with exit code 0)

then this in the logs,

"

2009-04-20 12:05:43.125 'PrimarySanProvider' 2508 info Return code for discoverLuns: 4
2009-04-20 12:05:43.125 'PrimarySanProvider' 2508 error The scripts returned an error, leaving the temporary file 'C:\WINDOWS\TEMP\vmware-SYSTEM\dr-sanprovider1'
2009-04-20 12:05:43.125 'PrimarySanProvider' 2508 error Unknown error encountered by the script
2009-04-20 12:05:43.125 'PrimarySanProvider' 2508 trivia Status of array 'array-800' is set to 'syncFailed'
2009-04-20 12:05:43.125 'PrimarySanProvider' 2508 warning Failed to re-sync with storage array: Unexpected MethodFault (dr.san.fault.ArrayUnknownFault) {
4 dynamicType = <unset>,
4 command = "discoverLuns",
4 msg = ""
4 }
"

Is this related to the issue you found ?? Also this thread mentions "install activeperl" . SRA 1.0.1 has a perl bin dir, all i did was added it to my path then it ran the perl commands.

RobertG

Hi depping,

Attached is the complete log bundle, I would greatly appreciate any help, I don't seem to have any authentication issues, i suspect it is the way the SVC repication is configured, would you have a sample SVC replication setup/config i could supply to the storage guys? they did tell me they create array groups that they add our ESX hosts too, looking at the new SRM & SVC SRA doco it states "vmware" must be added to the host object, not quite sure what this means.

RobertG

IBM Australia

Status update, adding "vmware" in the SVC host object seems to have solved my problem. A very interesting requirement

Thanks to all who assisted.

RobertG

All - I meant to update this yesterday!

We successfully ran a test and a failover with the most recent code, however, the code needs some tweaking as it has issues when there are more than 10 VM guests being recovered.

For those of you using the current code there is a bug related to how the SRA sees the volumes presented to the VC server. Essentially deleting a volume that contain a protected datastore may result in all the datastores dissapearing out of SRM until the volume is re-added.

cr