cancel
Showing results for 
Search instead for 
Did you mean: 

frozen service Service group offline after reboot

Dilipsun
Level 2
Certified

Hi

 

I have a service group named redsg running which is online on node1 and and offline  on node 2.

I froze resg and ran hastop -all -force ,and started it back again on both nodes

the service group state is online now on node1 and offline on node2 as earlier

But when i reboot the node the service group goes to offline on both nodes.

Any idea what makes this difference.

Thanks

dilip

2 ACCEPTED SOLUTIONS

Accepted Solutions

RiaanBadenhorst
Moderator
Moderator
Partner    VIP    Accredited Certified

As for the action taken before, hastop -all -force brings down only the cluster process (HAD) and leaves the clustered application in the exisiting state. Therefore when you restarted HAD with hastart, it found the application as you left it, online.

View solution in original post

mikebounds
Level 6
Partner Accredited

 A few points:

  1. If you freeze a service group, then when you reboot the node the frozen group is online on (uncleanly where rc scripts are not run) then the "Frozen" state prevents the service group from failing over to the alternate node, else the group would failover to another node.  
     
  2. If you reboot cleanly (where rc scripts ARE run), then the "Frozen" state will prevent VCS stopping (if the frozen group is online on the node being rebooting) and therefore the reboot will not complete (it will hang trying to stop VCS)
     
  3. When you start VCS, if a service is frozen, then the group will NOT autostart, else if the node VCS is being started on is in the AutoStartList, the group will online when VCS starts (if it is not already online on another node)
     
  4. If you are only rebooting one node at a time, then whether the group is frozen persistently or temporary, makes not difference.
     
  5. If you run hastop [-all | -local ] -force, then service groups are left in the state they are currently in.
     
  6. If you start VCS on a node where the resource component for a service group are already online on that node (i.e you stopped VCS when group was online using "-force"), then VCS sees the group is online when it starts, (whether it is frozen or not) and so does nothing to the group.

Hence when you reboot a node where the group is online and frozen, the frozen state prevents the group switching to other node when server goes down and prevents the group being started when the server comes back up.

Mike

View solution in original post

3 REPLIES 3

Gaurav_S
Moderator
Moderator
   VIP    Certified

Hello,

This would depend on what command you used to freeze the group ..  If you have used "persistent" flag in the "hagrp -freeze" command, the group will remain frozen & won't change the status once node is rebooted. So if a node is rebooted, as soon VCS is started, VCS will read the persistent frozen flag & will not take any action on group & hence it is offline.

Refer admin guide for more details

https://sort.symantec.com/public/documents/sfha/6.1/solaris/productguides/html/vcs_admin/ch06s12s07.htm

 

G

 

RiaanBadenhorst
Moderator
Moderator
Partner    VIP    Accredited Certified

As for the action taken before, hastop -all -force brings down only the cluster process (HAD) and leaves the clustered application in the exisiting state. Therefore when you restarted HAD with hastart, it found the application as you left it, online.

View solution in original post

mikebounds
Level 6
Partner Accredited

 A few points:

  1. If you freeze a service group, then when you reboot the node the frozen group is online on (uncleanly where rc scripts are not run) then the "Frozen" state prevents the service group from failing over to the alternate node, else the group would failover to another node.  
     
  2. If you reboot cleanly (where rc scripts ARE run), then the "Frozen" state will prevent VCS stopping (if the frozen group is online on the node being rebooting) and therefore the reboot will not complete (it will hang trying to stop VCS)
     
  3. When you start VCS, if a service is frozen, then the group will NOT autostart, else if the node VCS is being started on is in the AutoStartList, the group will online when VCS starts (if it is not already online on another node)
     
  4. If you are only rebooting one node at a time, then whether the group is frozen persistently or temporary, makes not difference.
     
  5. If you run hastop [-all | -local ] -force, then service groups are left in the state they are currently in.
     
  6. If you start VCS on a node where the resource component for a service group are already online on that node (i.e you stopped VCS when group was online using "-force"), then VCS sees the group is online when it starts, (whether it is frozen or not) and so does nothing to the group.

Hence when you reboot a node where the group is online and frozen, the frozen state prevents the group switching to other node when server goes down and prevents the group being started when the server comes back up.

Mike

View solution in original post