For more information regarding the security incident at F5, the actions we are taking to address it, and our ongoing efforts to protect our customers, click here.

Forum Discussion

COMMS-CORE_1795's avatar
COMMS-CORE_1795
Icon for Nimbostratus rankNimbostratus
Jul 29, 2015

Cluster resources reallocation

Hi,

 

I'm preparing a resource reallocation for a LTM cluster. I've been reading about, and there's a couple questions that I hope you can help me with.

 

The cluster is installed on two Viprion C2400 with one slot per Viprion, and each Viprion is hosting one member of the cluster. When this cluster were deployed, 2 cores per Viprion were assigned to each guest.

 

Now the active member of the cluster eventually runs out of memory, causing bad behavior on random services. I've been forcing a failover to free memory as a workaround. Since I still have 3 free cores on each Viprion, I'd like to increase the number of cores of each guest, from 2 to 4.

 

Since the cluster is in a "deployed" state, I'd like to know the best way to add more cores to both guests without affecting the service.

 

Would it be possible to...?:

 

1 - Standby guest: move from "deployed" to "provision" state. Reallocate with 4 cores. Then back to "deployed" state. Will be both guests from the cluster move to "provision" state this way (interrupting the service), or will the active one keep alive?

 

2 - If the active guest is not affected by the reallocation of the standby node, I will force a failover to the first one (which will be in a "deployed" state with 4 cores).

 

3 - Same procedure as in point 1 with the other guest and back to force a failover to it.

 

If you know a better way to do so or any reference about this issue (I haven't found any), please let me know.

 

Thanks a lot.

 

2 Replies

  • I've opened a support case, still no answer. If anyone has done something similar, please tell me. Thanks!
  • I guess your approach will be working, but I suspect stateful failover might fail for several connections when failing over between unequally (in terms of numbers of cores) equipped cluster partners.

     

    However, I'm also interested in the outcome of your support case. Would you please post the information here when you get the response?

     

    Concerning your question "Will be both guests from the cluster move to "provision" state this way (interrupting the service), or will the active one keep alive?": To my understanding, there is no way how changing the state of one vCMP guest on virpion1 might influence the state of the second vCMP guest on viprion2.