PCS HA Cluster questions

Discussion in 'Post-Deployment: Configuration and Setup' started by Machiasiaweb, Oct 6, 2015.

  1. Machiasiaweb

    Machiasiaweb Mega Poster

    Messages:
    167
    Hello,

    Some question about PCS HA Cluster is not clear after reading

    http://kb.odin.com/en/119207


    1) Can I just modify the config from one of cluster node or from MASTER node, and then distribute to all node?

    2) If (1) is not and require to modify each node. Should I also need adjust the

    VERSION=X to greater number?

    3) Need to restart any service after modify those parameters?

    4) What is means of 'drs' with parameter

    RESOURCE_RELOCATION_MODE=

    5) I tested with one of node crashed and the VPSs can be move to another node. Those VPS will be keep STOPPED and another node. However, when I checked by

    > shaman -c pcs top

    The VPS PWRR status is showing 'ON' and STATUS is 'ACTIVE'. Is there any something wrong config?

    Thanks!
     
  2. Pavel

    Pavel A.I. Auto-Responder Odin Team

    Messages:
    421
    Almost all these questions are well addressed in the documentation. Please check it beforehand.
    1) config is global
    2) - not relevant due to (1)
    3) no
    4) please refer to documentation
    http://download.cloudserver.odin.co.../Parallels_Cloud_Server_Users_Guide/37581.htm
    5) The question is not quite clear, I might misunderstand you here.
    Configuration seem to be correct. If that is the only question here - nothing needs to be done.
    However, if relocation does not work as expected I'd recommend you to contact technical support for deeper investigation
     
  3. Machiasiaweb

    Machiasiaweb Mega Poster

    Messages:
    167
    Hello,

    Thanks. For 1), 2) & 3) is cleared once modify from MASTER node, it will update all node as well.

    However, while further details check using -v stat. It showing following error. Did you know how to resolve it?

    ---
    # shaman -c pcsnode -v stat
    New version config found: 4, current 1. Please, ugrade the shaman package.
    Failed to read XML model: syntax error
    ----

    For 5), I guess is related to below
    http://kb.odin.com/en/124465

    However, my case is container. Not VM.

    *****************************
    New Question 6) when testing with iSCSI

    Example: I have two nodes

    192.168.29.204 (simulate crashed)
    192.168.29.206 (resource to be allocated)

    After 192.168.29.204 down
    ---
    # shaman -c pcsnode -v stat


    ISCSI ID STATE STATUS OWNER_IP OWNER_ID PRIORITY
    iqn.2014.06.com.parallels:store1 running Active 192.168.29.206 0ff06a1f313249c0 0
    ---
    ISCSI can be reallocate to 192.168.29.206

    But PORTAL IP is still keep using the old IP.

    --
    # pstorage-iscsi list
    IQN STATUS LUNs HOST PORTAL(s)
    iqn.2014.06.com.parallels:store1 running 2 0ff06a1f313249c0 192.168.29.204
    --

    Finally I need to manual change the PORTAL IP. Does it normal?


    Thanks!
     
    Last edited: Oct 6, 2015
  4. Pavel

    Pavel A.I. Auto-Responder Odin Team

    Messages:
    421
    iscsi IP address must not be the same as the hardware node. It's an absolutely separate IP address you assign.
    Obviously upon relocation it won't be changed, it was never designed to work this way.

    Use separate IP address for iscsi target.
     
    Last edited: Oct 9, 2015
    Machiasiaweb likes this.
  5. Machiasiaweb

    Machiasiaweb Mega Poster

    Messages:
    167
    Hello,

    Just new question that is what happen if others active node did not have enough license to get all VM/CT back to online?

    Thanks!
     
  6. Pavel

    Pavel A.I. Auto-Responder Odin Team

    Messages:
    421
    they will remain in "broken" state/pool until node is back online
    Relocation over the license boundaries won't happen
     

Share This Page