issue with MSCS

[Mohawk_Marvin]

Hi All

I have an issue with Microsoft Clustering and EV8 SP3.

My environment is as follows:
Active/Passive EV FSA server 2 node cluster, and targetting Celerra for the FSA targets.

So on my EV boxes I  need to install the FSA Agent and then run thru the FSA Cluster config to create my cluster resource. Done that and my resource is created.

So now the problem.....

My FSA resource will bring up the 3 services fine, however when choosing to take the FSA resource offline it changes the resource from Online to Offline instantly and looking in the Services console the 3 FSA services (placeholder, collector and blocking) remain running. This impacts me as when I move my group to the passive node I cannot shut down the Admin service (placeholder stays running so directory cant shut down).
Things of note are that my FSA resource does not have any dependencies other than the MSMQ, Data and IIS disks, so no mention of Admin or directory resource, this is odd as I know the placeholder needs these resources. Of course in the event of a total failure (like a reboot) the failover is perfect.

I have removed the agent and resource and reinstalled and recreated the resource still same issue.

There are no errors in eventlogs and a placeholder dtrace contains nothing of value.

Appreciate your thoughts

[JesusWept3]

I'm not a cluster or fsa guru at all, however only ev cluster will have a dependancy on msmq and iis, the fsa drivers and such will not as msmq is not used in file server archiving, especially on the file servers however since you are targeting a device as opposed to the file server I'm assuming the place holder service et al is installed on the ev server tbh though if there is nothing immediately obvious in the guides or technotes then your quickest resolution maybe to call support

[Mohawk_Marvin]

All the services install as the Agent gets installed no problem and retrieval etc works.
Call has been logged with Symantec however as yet they have not produced a solution or a step to progress with, so I thought I would try the forums :)

[Batmanfail]

Do you really need to install the FSA Agent on your EV Cluster?  I thought the Celerra DHSM API took care of the placeholder creation and recall via HTTP.   I think the storage service talks directly to the Celerra API.  Try removing the resource and see what happens...

[Mohawk_Marvin]

I do need the FSA Agent installed as I will want to configure FSA Reporting at some point soon, and unless I misunderstand things, I still need the FSA Agent installed somewhere regardless.

[Batmanfail]

You can install the FSA Agent on any Windows Server (I believe it's called an FSA Agent Proxy) so you don't have to install it on your EV cluster...

[Mohawk_Marvin]

From reading the Install guide:

"For NetApp and EMC Celerra file servers, the FSA Agent services that support placeholder shortcuts and FSA Reporting run on the Enterprise Vault Server."

I have had a quick poke around and all mention of FSA Agent Proxying say something along the lines of that quote, do you have a doc or guide I am missing showing configuration of the agent on another NTFS box other than the EV ones?

[Batmanfail]

True.  My bad! 

Following excerpt taken from Reporting.pdf (pg64);

"For NetApp and EMC Celerra file servers, an Enterprise Vault server acts as a proxy server for performing the FSA Reporting file collection."

...so, back to your FSA resource issue then...

I take it this is a Windows 2003 MSCS (2 x node Active/Passive)?

[Mohawk_Marvin]

It is indeed Active/Passive 2003 MSCS.

And the FSA Resource will happily bring the 3 File Services online however, if I elect to shut down the FSA resource, it doesnt take them offline, so when trying to shut down the Admin resource, it cant due to the Placeholder service running. Therefore when the admin service fails to go offline in time... it fails and cluster fails over.

So what I need/want is for the resource to actually govern the 3 services correctly :)

As per my first post, I have uninstalled the FSA agent, rebooted, installed the agent on both nodes, rebooted, then run the cluster config to create my resource. All works fine to this point. Then we get back to the issue where the resource does not shut down the 3 services.

[Batmanfail]

Did you configure the FSA target with the EV Server Network Name Resource?  If you added the target as one of the nodes or as the cluster name, maybe this is the issue?  Try the target name as the EV Server Network Name Resource instead...

Also, something to note is that the Enterprise Vault FSA Reporting Web service is not supported on Windows Server 2003 x64 edition...

[Mohawk_Marvin]

Good thinking Batman, sorry been desperate to say it all day.

Already tried that as well it needs to match the rest of my resource names otherwise it does not work at all.

Dont forget.. at the moment my FSA Resource can bring the 3 Fil Services online no problem, it just cant take them down even though the resource shows offline.

[Liam_Finn1]

OK this may seem like a silly question and I'm sure you have checked this but....are the services set to manual and are the services set to be dependent on the resource disk so when the disk moves they are forced offline and then come on-line on the passive node?

[Batmanfail]

- Only "joshing"...Mr BaBa

If it's any consolation, I installed the FSA agent on my cluster and have the same issue.  If I take the resource offline the services remain started.  However, if I manually stop the services and then use the "Bring Online" option to start them, they start fine.   BTW i'm running 8SP3 on Win2k3 R2 SP2 EV clustered...

Now, that reminds me, I must get my hair cut.....at the barbers, hohoho

[Batmanfail]

Just tried to do a "Move Group" and like you say, the 3 FSA services remain started (even if set to manual and dependent on physical disks).  Therefore the Admin service can't stop and the cluster resource group remains in a Pending state.  If you manually stop the FSA services (placeholder inparticular) the resource group move completes successfully, but I had to also manually bring the FSA resource online because it had failed!

Not the most elegant resource failover.  Maybe this one has been missed by the EV QA team...?

[BravoZulu · Accepted solution]

Hi Josh,

I have a workaround for this issue, but bear in-mind that it may not be suitable for your customer.  Here's what I do;

1.  On the Dependencies tab of the "<ServerName> Enterprise Vault FSA" resource, include the "Enterprise Vault Directory Service".  This is important when the resource tries to come online (so it doesn't fail).

2.  Create a batch file (i.e. stopFSAservices.bat) on your c drive with the following line;
net stop "Enterprise Vault File Placeholder Service"

3.  Create an event trigger on your EV cluster which responds to an attempted failover (as follows);

eventtriggers /create /l System /eid 1203 /t INFORMATION /tr TriggerName /ru evserviceaccount@yourdomain.com /rp password /tk "c:\stopFSAservices.bat"

NOTE:
You will have to do this on both cluster nodes!

Hope that helps with the issue?

BZ

[Batmanfail]

Works for me BZ!  Looks like you have a solution Ali BaBa...

[Mohawk_Marvin]

Its not a solution more a workaround and I kind of had something in place to make this work, I was hoping I had missed something but it seems this is a "bug" or feature. Glad that SYMC support now have this.

Thanks for your time Batman and BZ