TOPIC:

Hi

I'm evaluating HA-Lizard with no-san installer and XEN Server 6.5

I successfully installed it in a nested env, everything works there.

Now I try the same thing on two identical HP DL585 with AMD Opterons. For replication I use two NetXen NX3031 1Gbit Adapters. And there happens the first problem. When i changed the roles the adapter stopped responding, same on rebooting. Sometimes it failes sometimes not. So changed the cables and tried several reboots, and the problem with the adapter went away.


Now im trying to change the roles again: iscsi-ha changes the role quickly (xapi count 2 and xapi delay 3). Xen also does, but after the role change the vm is not more responding and hangs with timeout on disk. If i try to restart the maschine on the new master it says no bootable device. iscsi-cfg says tgt running. I still can boot the VM on the former master which is the slave now. iscsi-ha shows stopped so i don't really get it. Any help would be appreciated. both can ping the floating IP

I had the same Issue with a fully patched xen. Now i don't have any updates installed.

Before i changed roles i could start vms on both Hosts.

Thanks in Advance!

Yours Sincerely,
Joel

Can you check if OVS is enabled
cat /etc/xensource/network.conf
The result should be "bridge"

OVS is known to have issues with DRBD which results in the interface not responding.

Both Hosts output bridge. If i change the role back to the former master it working again: both hosts can start vms from the iscsi lun. live migration works as well.

if slave is master: only new slave can start vms. drbd master role is on the new master. So it seems the slave has issues connect to iscsi lun if the drbd/xen master role is local. When i then try to migrate from new slave the new master it migrates and end up with successfull but the vm hangs.

check the attachment this is what i get, when i try to start a vm on xen2(slave) when its the master.

Oh i dont think you can read the error, it got smaller.
I've resized it. check attachement

So there is no problem with the network?

What else i did, i had problems with local storage that should be created during installation. So i skipped to choose a disk for the vm repository. So there where two 4gb partition from xen, i've created an lvm partition with gdisk on both hosts /dev/sda3 with 1.85tb. which i then used for the ha-lizard setup. Is this correct?

In the video you tell also you have fixes for that problem with no local storage after installation?

there is also some spam errors if i try to write mails from the ha-lizard webpage to get professional service.

hmm the pv directory is different, on my nested env they have the same? also there is another xen storage on xen1? Maybe its really this strange installation behaviour on storage creation, thats always different. I thought i could avoid this by creating the vm repository later.

HP ProLiant DL585 Gen7 with 1 logical drive (raid 10, 4 drives)
raid controller: 410i 1gb


[root@xen1 ~]# pvscan
PV /dev/sdc VG VG_XenStorage-793837fd-3565-5771-b901-79db1e04ee7b lvm2 [1.81 TB / 1.78 TB free]
PV /dev/sda2 VG VG_XenStorage-9add7ca7-2627-e998-e907-9a8e337e7eaa lvm2 [3.99 GB / 3.98 GB free]
Total: 2 [1.82 TB] / in use: 2 [1.82 TB] / in no VG: 0 [0 ]

[root@xen2 ~]# pvscan
PV /dev/sdb VG VG_XenStorage-793837fd-3565-5771-b901-79db1e04ee7b lvm2 [1.81 TB / 1.78 TB free]
Total: 1 [1.81 TB] / in use: 1 [1.81 TB] / in no VG: 0 [0 ]