digiturf.net

Freezing and Unfreesing Service Groups in Veritas cluster ( VCS )


og-dcc01a # hastatus -summ
                                                – The above command give many details about cluster that is running on group of servers. The information that can be found / understood from it is as follows.

     * If you see “cvm” (Cluster Volume Manager) in output, it means that, the type of cluster that is running is “Parallel Cluster” not the fail-over one.
     * You can see No.of Hosts that are part of cluster under “SYSTEM STATE” menu.
     * It shows, service groups names and other information related to them that are setup on it. (All the others except “CluterService”, “cvm” are service groups that we generally work with…!)

og-dcc01a # haconf -makerw     – Before freezing service groups, we should make VCS cluster config RW.
og-dcc01a # hagrp -freeze cmddsdb_grp -persistent  
og-dcc01a # haconf -dump -makero    – make “VCS Config” RO, once freezing completes.

* -persistent enables the freeze / unfreese to be remembered when the cluster is rebooted. Config
   must be “RW” to use this option.
* You can see from below output that service group “cmddsdb_grp” is shown in the list of “
   GROUPS FROZEN”.

og-dcc01a # hastatus -summ

— SYSTEM STATE
— System               State                Frozen

A  og-dcc01a      RUNNING              0
A  og-dcc01b      RUNNING              0

— GROUP STATE
— Group                  System           Probed     AutoDisabled    State

B  ClusterService  og-dcc01a           Y               N               OFFLINE
B  ClusterService  og-dcc01b           Y               N               ONLINE
B  cmddsdb_grp    og-dcc01a           Y               N               ONLINE
B  cmddsdb_grp    og-dcc01b           Y               N               ONLINE
B  cvm                   og-dcc01a           Y               N               ONLINE
B  cvm                   og-dcc01b           Y               N               ONLINE

— GROUPS FROZEN
— Group

C  cmddsdb_grp

— RESOURCES DISABLED
— Group                      Type                Resource

H  cmddsdb_grp     CFSMount        cmddsdb_mnt1
H  cmddsdb_grp     CFSMount        cmddsdb_mnt2
H  cmddsdb_grp     CFSMount        cmddsdb_mnt3
H  cmddsdb_grp     CVMVolDg      cmddsdb_voldg1
H  cmddsdb_grp     CVMVolDg      cmddsdb_voldg2
H  cmddsdb_grp     CVMVolDg      cmddsdb_voldg3
H  cmddsdb_grp     Oracle               cmddsdb

                * As per the above output we can find that, it is 2 Node “Parellel cluster”, we can freeze the service group on both the systems by logging into them at same time.
                * Once the maintenance activity is complete, we can “unfreeze” the service groups by using following method.

og-dcc01a # haconf -makerw     – 1st make VCS Config RW.
og-dcc01a # hagrp -unfreeze cmddsdb_grp -persistent    – Unfreeze the service group(s).
og-dcc01a # haconf -dump -makero      – Once unfreeze is done, make VCS config RO.

## You get the following error when “-persistent” is not given while trying to freeze / unfreeze SG.
og-dcc01a # hagrp -unfreeze cmddsdb_grp      
VCS WARNING V-16-1-40201 Group is not temporarily frozen

* Now from cluster summery, you can see that SG “cmddsdb_grp” is gone from “GROUP STATE”
   menu, which means it is no more in freeze state. We also see that, the same SG is in “ONLINE”
   state in both the nodes in cluster because it is “Parallel cluster” as i’ve mentioned earlier.
* Here you see that, these steps are executed on “og-dcc01a”, the same steps are also executed at
   same time on “og-dcc01b”.

og-dcc01a # hastatus -summ

— SYSTEM STATE
— System               State                Frozen

A  og-dcc01a      RUNNING              0
A  og-dcc01b      RUNNING              0

— GROUP STATE
— Group                  System           Probed     AutoDisabled    State

B  ClusterService  og-dcc01a         Y               N               OFFLINE
B  ClusterService  og-dcc01b         Y               N               ONLINE
B  cmddsdb_grp    og-dcc01a         Y               N               ONLINE
B  cmddsdb_grp    og-dcc01b         Y               N               ONLINE
B  cvm                   og-dcc01a         Y               N               ONLINE
B  cvm                   og-dcc01b         Y               N               ONLINE

* Parallel    :   Service Groups will be running on all the nodes which are part of cluster at a time.
* Fail-over  :   SG’s will be running on one server at a time, on rest of them they will not be running.

Leave a Reply