Visit Open-E website
Results 1 to 3 of 3

Thread: Multipath Failover with 2 Switches

  1. #1

    Default Multipath Failover with 2 Switches

    Hello,

    i have use this How-To "Open-E DSS V6 How to setup iSCSI Failover with XenServer" with open-e 6.0up95.xxxx.6156 (full) to configure a open-e failover solution with multipath.

    My Config:

    2 Dell Powerconnect 5524 with Stacking 10GB/s.

    Ping Nodes: 192.168.2.106 and 192.168.3.106 -> Switch 1
    Ping Nodes: 192.168.2.107 and 192.168.3.107 -> Switch 2

    Storage Client 1 (Virtualization Host;multipath): 192.168.10.100 -> Switch 1 ; 192.168.20.100 -> Switch 2
    Storage Client 2 (Virtualization Host;multipath): 192.168.10.200 -> Switch 1 ; 192.168.20.200 -> Switch 2

    Open-E VSS Node 1:
    eth0 -> Disabled
    eth1 -> only for Web Gui
    eth2 -> 192.168.1.230 -> Switch 1 (Auxiliary connection)
    eth3 -> 192.168.2.230 -> Switch 1 (Auxiliary connection and Virtual IP 192.168.10.250)
    eth4 -> 192.168.3.230 -> Switch 2 (Auxiliary connection and Virtual IP 192.168.20.250)
    eth5 -> 192.168.4.230 -> Direct with Crossover (Volume Replication)

    Open-E VSS Node 2:
    eth0 -> Disabled
    eth1 -> only for Web Gui
    eth2 -> 192.168.1.240 -> Switch 2 (Auxiliary connection)
    eth3 -> 192.168.2.240 -> Switch 2 (Auxiliary connection and Virtual IP 192.168.10.250)
    eth4 -> 192.168.3.240 -> Switch 1 (Auxiliary connection and Virtual IP 192.168.20.250)
    eth5 -> 192.168.4.240 -> Direct with Crossover (Volume Replication)


    But i have follow problems:

    - If Primary node is active, so is all ok, but if a make a manual failover so thet performance is terrible and i can't connect to a virtal machine.
    - Whan i put all connection to the switch 1, so is all ok.

    But why is this so? And what is whan a Switch go down?

    The Virtual IP 192.168.10.250 of Primary Node is on Switch 1 and both Client IP 192.168.10.100/192.168.10.200 too.
    The Virtual IP 192.168.20.250 of Primary Node is on Switch 2 and both Client IP 192.168.20.100/192.168.20.200 too.

    In this Scenario is all ok, but whan The Secondary node is active so is follow configurated:

    The Virtual IP 192.168.10.250 of Secondary Node is on Switch 2 and both Client IP 192.168.10.100/192.168.10.200 on Switch 1.
    The Virtual IP 192.168.20.250 of Secondary Node is on Switch 1 and both Client IP 192.168.20.100/192.168.20.200 on Switch 2.

    In this Scenario i have the problems.

    But what is whan a Switch go down? So is only a 192.168.10.250 or 192.168.20.250 online?

    I hope for help.

  2. #2

    Default

    so, i have testing the other stacking port and that was more better. At the moment i think that a Switch have a problem with one of the stacking ports. Tomorow i become a another switch, so that i can make more tests.

    But a another Question:

    I have install Open-E VSS6 up90 and this have a better performance as the up95 version.

    Iometer in a VM on up95:
    I/Os / second = ca. 6000
    Total MB / second = 96MB/s

    Iometer in a VM on up90:
    I/Os / second = ca. 8000
    Total MB / second = 125MB/s

    Why is there so a big difference?

    Is as posible to optimate up95?

    EDIT:
    Is it normal, whan a switch go down (in multipath configuration) that i must wait 2 minutes for a normal work?

    thanks.
    Last edited by michi616; 05-07-2012 at 06:38 PM.

  3. #3
    Join Date
    Aug 2008
    Posts
    236

    Default

    Quote Originally Posted by michi616 View Post
    EDIT:
    Is it normal, whan a switch go down (in multipath configuration) that i must wait 2 minutes for a normal work?

    thanks.
    Check your spanning tree configuration on your ports. Make sure you turn on portfast or if you are using RSTP, that you make your ports edge ports.
    You can also adjust the iscsi client timeouts as well. This can improve the speed of recovery when you loose a link.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •