Where did my service group Failover to in a 3 node cluster?

[Gurminder_Moh1]

I know you can look into the engine logs to determine this, but I am trying to find scripted/automated ways to send my group an email when a SG fails over to any of the nodes in a VCS cluster.

So please let me know what you do to manage this in your current environment. I have mutiple 3 node clusters in my organization and each node can run upto 10-15 oracle databases. In case of a node crash, it has become painful to determine where did each of my service group failed over to except for looking through logs (which is time consuming), hence need an automated way.

Any and all ideas are welcome.

[Gaurav_S]

Have you read about VCS SMTP/SNMP notifications ?

Have a look at VCS users guide or VCS admin guide (if 5.1 version)... & check for above mentioned section...

users guide can be found:

https://vos.symantec.com/documents


Gaurav

[Anoop_Kumar1 · Accepted solution]

You can add and configure a notifier resource in cluster to send the VCS messages to SNMP and SMTP servers.

Please read about NotifierMngr agent in VCS bundled agent guide for more details.

http://sfdoccentral.symantec.com/sf/5.0/solaris/pdf/vcs_bundled_agents.pdf

[Gurminder_Moh1]

Thanks for your suggestions Anoop and Gaurav. I Will surely look up the documents you  guys are referring to.

We do use notifier resoruce and have SMTP notifications setup in our environment. We do get emails on when a service groups faults or a resrouce faults, but I have yet to see an email or notification that tells which node my SG failed over to. Maybe we are missing something.

I will go though the documents as well to determine what we are missing, but if you have any specific example, that will be great.

group Notifier (
    SystemList = { xxx = 0, yyy = 1, zzz = 2 }
    AutoStartList = { xxx, yyy, zzz }
    )

    NotifierMngr Email (
        SmtpServer = "aaa.example.com"
        SmtpRecipients = { "abc@example.com" = Error }
        )

  

[D_White1]

Hi Gurminder,

Without looking through the docs, my guess is that the completion of group online would be classified as an "Information" level message or whatever the correct classification is.  Your configuration above suggests that you are sending "Error" level messages only so you won't receive a notification about where a group comes online (service group coming online is not an error but a service group faulting is an error so you'll receive emails about faults but not online events).

If you change the configuration to send emails for all events, this would probably give you what you need but may be overkill?

Alternatively, you could consider implementing postonline triggers which would allow you to run custom scripts (such as send a mail) when the group comes online?

[EJameson]

Hello Gurminder,

As you will have seen from a brief review of the documentation, the VCS Notifier should be exactly what you'll need to monitor the state and events taking place of your cluster, regardless of which node it is currently online with. Ultimately, the Notifier service group can be configured through the Veritas Cluster Configuration Wizard (VCW).  This can be done to existing cluster, or as part of a new cluster configuration. In your current cluster configuration(s), simply:

1) Click start and select the Cluster Configuration Wizard from the Symantec Storage Foundation Start Menu group

2) Go through the wizard inserting the correct information for your existing cluster, while choosing the option to Edit An Existing Cluster

3) At the last section of the wizard, you should see the option to place a checkmark next to the option for creating a Notifier Service Group in the cluster. Choose this option and hit next.

4) You'll now need to enter the proper SNMP / SMTP information and adjust the Attribute values to specify whether you wish to receive contact regarding complete failures or resources coming online / offline.

Additionally, you can setup notification emails through the Volume Manager portion of SFW, though this will only provide alert contact when having issues within the Volume Manager, as opposed to the cluster; however, if there are issues with disks or Disk Groups, this will allow for monitoring those behaviors as well.

The link provided by Gaurav in the first post should provide you with access to all of the available documentation for Storage Foundation on each supported platform and as such, I would suggest reviewing the Admin Guide for the Cluster for indepth configuration information for the Notifier attributes.

I hope this has provided at least a bit more help than the folks here have already offered, but if you have any questions or concerns, please contact us.

Thanks,

E. Jameson