HPE Community read-only access December 15, 2018
This is a maintenance upgrade. You will be able to read articles and posts, but not post or reply.
Hours:
Dec 15, 4:00 am to 10:00 am UTC
Dec 14, 10:00 pm CST to Dec 15, 4:00 am CST
Dec 14, 8:00 pm PST to Dec 15, 2:00 am PST
cancel
Showing results for 
Search instead for 
Did you mean: 

2+1 cluster failover

 
asce
Occasional Advisor

2+1 cluster failover

I have 2+1 cluster.one node always active to other.Now ,want to set pkg failover policy is failover to active node.

How I can do it.Thanks!
11 REPLIES
sujit kumar singh
Honored Contributor

Re: 2+1 cluster failover

Hi

If this an MCSG that is Service Guard cluster check the type of the package in the #cmviewcl -v

U can see well that the package shall be of the type Multinode or System Multinode and that the package is not of the type failover.

Only packcages of the type failover can be made to failover .

Those with the system multinode or multinode are generally cannot be configured to failover.


Regards
Sujit

u can post the #cmviewcl -v



output, that will be veery helpful

Regards

Sujit
asce
Occasional Advisor

Re: 2+1 cluster failover

Hiï¼

this is plan,So I can not display cmviewcl.
cluster have 3 nodes.
node1 have a package,node2 have a package.node3 free.
set package always failover to free node.not switch back.I don't know set it.
sujit kumar singh
Honored Contributor

Re: 2+1 cluster failover

Hi,

Assuming that three nodes are node1, node2 and node3 and the packages are pkg1 and pkg2.
in normal conditions pkg1 runs on node 1 and pkg2 runs on the node2 and node3 does not run any package but acts as the standby node for both the packages.

for this edit the pkg1 config file say pkg1.conf as follows:

PACKAGE_NAME pkg1
PACKAGE_TYPE FAILOVER
FAILOVER_POLICY CONFIGURED_NODE
FAILBACK_POLICY MANUAL
NODE_NAME node1
NODE_NAME node3

and for the package pkg2 edit the config file pkg.conf as

PACKAGE_NAME pkg2
PACKAGE_TYPE FAILOVER
FAILOVER_POLICY CONFIGURED_NODE
FAILBACK_POLICY MANUAL
NODE_NAME node2
NODE_NAME node3


Check and apply the conf file with the other necessary things for the package.


U will get the packages work in the way u wish.

Regards
Sujit
asce
Occasional Advisor

Re: 2+1 cluster failover

Thanks Sujit.

for the action.then node1 failover to node3 or node2 failover node3.after rebuild,node1 or node2 is active.for next failover.How can config package auto switch to active node?

Regards,
Liqiang
sujit kumar singh
Honored Contributor

Re: 2+1 cluster failover

hi

what u wish to achieve is that u want the failback of the package to be automatic that is when the actual node comes up the package shall auto switch back or failback to the primary node.

This is done using the pgf conf file editing for the desired package as

and for the package pkg2 edit the config file pkg.conf as

PACKAGE_NAME pkg2
PACKAGE_TYPE FAILOVER
FAILOVER_POLICY CONFIGURED_NODE
FAILBACK_POLICY MANUAL
NODE_NAME node2
NODE_NAME node3


Regards
Sujit
sujit kumar singh
Honored Contributor

Re: 2+1 cluster failover

There is a correction in the earlier post: Apoligize


If u wish to achieve is that u want the failback of the package to be automatic that is when the actual node comes up the package shall auto switch back or failback to the primary node.

This is done using the pgf conf file editing for the desired package as

and for the package pkg2 edit the config file pkg.conf as

PACKAGE_NAME pkg2
PACKAGE_TYPE FAILOVER
FAILOVER_POLICY CONFIGURED_NODE
FAILBACK_POLICY AUTOMATIC
NODE_NAME node2
NODE_NAME node3


Regards
sujit kumar singh
Honored Contributor

Re: 2+1 cluster failover

Hi,


One word of caution. I have rarely seen the systems in the production havibg the FAILBACK_POLICY set as AUTOMATIC for the packages. The basic idea is that why FAILBACK_POLICY is generally set to manual is that:
1)If due to some transient reason the Primary node failed, the package will fsil to the adaptive node, that is OK, but say whwn the Primary node comes up again then the package will try automatically to move to the Primary node if AUTOMATIC FAILBACK_POLICY is configured. The node though might have come back up but it might not be capable actually to handle the pacakge. Better that we can plan the failback manual once we have confirmed the reason of the Primary node failing , analysis done and we are sure that the node is healthy enought to again run the package.

2) Also When the Pry node had failed this had caused the unforesaid downtime of the package but when it comes up package might again get down on the adaptive node to start on the Pry node if FAILBACK_POLICY is AUTOMATIC. This might lead to unnecessary switching of the packcge in the business hours. This Failback if set manual we can later on switch the package to the Primary node during off-peak business hours.


Regards
Sujit
asce
Occasional Advisor

Re: 2+1 cluster failover

Hiï¼

thanks your adviceã
for the failure node,will join the cluster after repair .
whether is it possible if config pkg.conf to
FAILOVER_POLICY MIN_PACKAGE_NODE.

Regardsï¼
Liqiang
sujit kumar singh
Honored Contributor

Re: 2+1 cluster failover

Hi,

FAILOVER_POLICY MIN_PACKAGE_NODE means that once the primary node where the package is running if that pry node fails then the Package shall failover to the available node in the cluster that has the least number of packages running on it.

If u set FAILBACK_POLICY as AUTOMATIC then the package tries to switch back to the original node automatically when the Riginal node comes up and joins the cluster or u make that join the cluster.


Regards
Sujit


asce
Occasional Advisor

Re: 2+1 cluster failover

Hi,

I will carry on some test, having results will be replied to you again. Thank you for your help again.

Regards,
Liqiang
sujit kumar singh
Honored Contributor

Re: 2+1 cluster failover

Hi Sir,


Thanks for thanking .... but u can also assign some marks for the replies even if u are not finding them helpful.

No Probs ... please be back with ur findings .. Always welcome Sir ..

Regards
Sujit