run HA with two nodes

rickygm · Feb 17, 2016

Hi forum , Hi have two nodes running in a cluster with Proxmox 3.4, I'm making run HA for vm, but I still have not succeeded, if I shutdown a node , in the cluster the virtual machine is migrated to another node, but does not start, remains off.

ilo4 the administration is correct, I can connect perfectly
My server are Hp DL380P

<?xml version="1.0"?>
<cluster config_version="17" name="RPI">
<cman expected_votes="1" keyfile="/var/lib/pve-cluster/corosync.authkey"/>
<fencedevices>
<fencedevice agent="fence_ilo4" ipaddr="192.168.21.12" login="Administrator" name="fencea" passwd="XXXXX"/>
<fencedevice agent="fence_ilo4" ipaddr="192.168.21.11" login="Administrator" name="fenceb" passwd="XXXX1"/>
</fencedevices>
<clusternodes>
<clusternode name="hyper1" nodeid="1" votes="1">
<fence>
<method name="1">
<device action="reboot" name="fencea"/>
</method>
</fence>
</clusternode>
<clusternode name="hyper2" nodeid="2" votes="1">
<fence>
<method name="1">
<device action="reboot" name="fenceb"/>
</method>
</fence>
</clusternode>
</clusternodes>
<rm>
<pvevm autostart="1" vmid="5000"/>
</rm>
</cluster>

rickygm · Feb 18, 2016

Hi, I'm doing the procedure to simulate a fall of a node is shutdown give the tty one of them.

adamb · Feb 18, 2016

rickygm said:
Hi forum , Hi have two nodes running in a cluster with Proxmox 3.4, I'm making run HA for vm, but I still have not succeeded, if I shutdown a node , in the cluster the virtual machine is migrated to another node, but does not start, remains off.

ilo4 the administration is correct, I can connect perfectly
My server are Hp DL380P

<?xml version="1.0"?>
<cluster config_version="17" name="RPI">
<cman expected_votes="1" keyfile="/var/lib/pve-cluster/corosync.authkey"/>
<fencedevices>
<fencedevice agent="fence_ilo4" ipaddr="192.168.21.12" login="Administrator" name="fencea" passwd="XXXXX"/>
<fencedevice agent="fence_ilo4" ipaddr="192.168.21.11" login="Administrator" name="fenceb" passwd="XXXX1"/>
</fencedevices>
<clusternodes>
<clusternode name="hyper1" nodeid="1" votes="1">
<fence>
<method name="1">
<device action="reboot" name="fencea"/>
</method>
</fence>
</clusternode>
<clusternode name="hyper2" nodeid="2" votes="1">
<fence>
<method name="1">
<device action="reboot" name="fenceb"/>
</method>
</fence>
</clusternode>
</clusternodes>
<rm>
<pvevm autostart="1" vmid="5000"/>
</rm>
</cluster>

The only way a 2 node cluster can work in 3.4 is with a quorum disk. Are you using a quorum disk? Based on your results I am guessing you aren't. You have to maintain 2 votes for quorum. If 1 server goes down, there is only 1 vote and no way to achieve quorum other than manually setting it. Quorum disks aren't supported in proxmox 4. I would suggest adding a 3rd node and it will make life much easier. It can be a very light weight node, it just needs to provide a quorum vote.

rickygm · Feb 18, 2016

adamb said:
The only way a 2 node cluster can work in 3.4 is with a quorum disk. Are you using a quorum disk? Based on your results I am guessing you aren't. You have to maintain 2 votes for quorum. If 1 server goes down, there is only 1 vote and no way to achieve quorum other than manually setting it. Quorum disks aren't supported in proxmox 4. I would suggest adding a 3rd node and it will make life much easier. It can be a very light weight node, it just needs to provide a quorum vote.

Adamb Hi , right now I'm using Proxmox 3.4, a third node is a little complicated by issues of money, but what would be the easiest route, have a quorum disk.

I'm using a Hp Eva6300, you could use some volume to create the quorum disk?

adamb · Feb 18, 2016

rickygm said:
Adamb Hi , right now I'm using Proxmox 3.4, a third node is a little complicated by issues of money, but what would be the easiest route, have a quorum disk.

I'm using a Hp Eva6300, you could use some volume to create the quorum disk?

Yep in theory you could present a new disk from the HP Eva6300 and use that as quorum (Only needs to be like 10-20MB). Its definitely not the best design but will atleast get you to the point of a basic HA setup.

rickygm · Feb 18, 2016

adamb said:
Yep in theory you could present a new disk from the HP Eva6300 and use that as quorum (Only needs to be like 10-20MB). Its definitely not the best design but will atleast get you to the point of a basic HA setup.

I have a question, see if you can give me some idea, I have 4 disks submitted to both nodes in the cluster, these and have used and created LVM volumes, yet I have free space, but not if I can be useful? Since each disc is already using

look:

lvm> vgs
VG #PV #LV #SN Attr VSize VFree
lvm_DFAST1 1 3 0 wz--n- 730.00g 325.00g
lvm_DFAST2 1 1 0 wz--n- 730.00g 570.00g
lvm_DLOW1 1 1 0 wz--n- 4.00t 3.93t
lvm_DLOW2 1 0 0 wz--n- 4.29t 4.29t
pve 1 4 0 wz--n- 279.24g 3.24g

now I have a Synology NAS NFS added to the cluster, for backup of my vm, but then have a free disk NAS 4TB I think it could use for iSCSI

http://i1132.photobucket.com/albums/m579/rickygm/synologi_zpsageuwtpn.png

What could be the best scenario?

adamb · Feb 18, 2016

rickygm said:
I have a question, see if you can give me some idea, I have 4 disks submitted to both nodes in the cluster, these and have used and created LVM volumes, yet I have free space, but not if I can be useful? Since each disc is already using

look:

lvm> vgs
VG #PV #LV #SN Attr VSize VFree
lvm_DFAST1 1 3 0 wz--n- 730.00g 325.00g
lvm_DFAST2 1 1 0 wz--n- 730.00g 570.00g
lvm_DLOW1 1 1 0 wz--n- 4.00t 3.93t
lvm_DLOW2 1 0 0 wz--n- 4.29t 4.29t
pve 1 4 0 wz--n- 279.24g 3.24g

now I have a Synology NAS NFS added to the cluster, for backup of my vm, but then have a free disk NAS 4TB I think it could use for iSCSI

http://i1132.photobucket.com/albums/m579/rickygm/synologi_zpsageuwtpn.png

What could be the best scenario?

Yea I would present that 4TB disk or a small piece of it over iscsi to both hosts, then setup the quorum disk on that.

rickygm · Feb 18, 2016

let me do it, and put progress

rickygm · Feb 22, 2016

Hi, i have a doubt, when I add a SCSI connection, if I reboot a node of the cluster and want to see on the other node iSCSI storage, often you show me this error message:
connect timed out failed
the connection gets slow and I have to remove the datastore option.

any idea? , where I see the log.

adamb · Feb 22, 2016

rickygm said:
Hi, i have a doubt, when I add a SCSI connection, if I reboot a node of the cluster and want to see on the other node iSCSI storage, often you show me this error message:
connect timed out failed
the connection gets slow and I have to remove the datastore option.

any idea? , where I see the log.

Now that you have a qurorum disk, can you provide the output of "clustat" from each node?

rickygm · Feb 22, 2016

Hi adamb , I'm having trouble when I add the disk iscsi, only what I see on one of the nodes, I think should appear in both.

Node1
Disk /dev/sdr: 10.7 GB, 10737418240 bytes
64 heads, 32 sectors/track, 10240 cylinders, total 20971520 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 8192 bytes / 8192 bytes
Disk identifier: 0x00000000

look link
http://i1132.photobucket.com/albums/m579/rickygm/node1_zpskxz0qael.png

Node2
not show disk on cli interface
look link
http://i1132.photobucket.com/albums/m579/rickygm/node2_zpsovuzhwvk.png

adamb · Feb 22, 2016

rickygm said:
Hi adamb , I'm having trouble when I add the disk iscsi, only what I see on one of the nodes, I think should appear in both.

Node1
Disk /dev/sdr: 10.7 GB, 10737418240 bytes
64 heads, 32 sectors/track, 10240 cylinders, total 20971520 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 8192 bytes / 8192 bytes
Disk identifier: 0x00000000

Node2
not show disk

How are you logging into the iscsi target? From the GUI or using iscsiadm?

rickygm · Feb 22, 2016

I use iscsi target from GUI.

adamb · Feb 22, 2016

rickygm said:
I use iscsi target from GUI.

Honestly not sure if that logs both hosts into the target. Does your central storage allow multiple logins from multiple initiators?

I would just log into the target using iscsiadm from each host.

rickygm · Feb 22, 2016

I'm using a synology disco, shows me that both servers are connected
http://i1132.photobucket.com/albums/m579/rickygm/disk_iscsi_zpswysexkud.png

adamb · Feb 22, 2016

rickygm said:
I'm using a synology disco, shows me that both servers are connected
http://i1132.photobucket.com/albums/m579/rickygm/disk_iscsi_zpswysexkud.png

Not sure, it does seem like you are logged in.

What is the output of "iscsiadm -m session" on both front ends?

rickygm · Feb 22, 2016

Strange Behavior, view output on both nodes

http://i1132.photobucket.com/albums/m579/rickygm/output_zpsikxqaod7.png

now i can acces on node2 a iscsi disk

adamb · Feb 22, 2016

Yea that is odd. On a side note, one of your hosts is logged into the iscsi target twice. The other is logged in only once.

rickygm · Feb 22, 2016

I have a little afraid, put the quorum disk with this behavior.

adamb · Feb 22, 2016

rickygm said:
I have a little afraid, put the quorum disk with this behavior.

Personally I would log out of the taget and then log back in with iscsiadm. Ensure you only have 1 path per initiator and then move forward.

run HA with two nodes

Renowned Member

Renowned Member

Famous Member

Renowned Member

Famous Member

Renowned Member

Famous Member

Renowned Member

Renowned Member

Famous Member

Renowned Member

Famous Member

Renowned Member

Famous Member

Renowned Member

Famous Member

Renowned Member

Famous Member

Renowned Member

Famous Member