Hello all.
I have a 3 node cluster HA working with idrac/IPMI fencing. It is working when some node has a problem.
At this moment, for testing, 3 nodes are connected to one switch and I'm testing what happens if all is working and suddenly switch is powered off.
This is the state of the cluster before removing power to the switch:
PVEVM has 192.168.1.144 and ping it's working:
At this point I cut the power to the switch and wait for some minute and power on again the switch.
The new info on clustat is this:
And pvevm 100 not starts again,ping is not possible host because is not accesible. I can't start manually, too. Is not possible that VM goes online when the switch is power on again? .
At this point is not possible to restart cman:
Not possible restart rgmanager, it hangs on:
Is necessary to restart all nodes to get the inicial config and pvevm running?
Please, someone can help me How I can restart all services withouth restart nodes and how to pvevm goes online automated when switch is turn on again?
Thank you all.
I have a 3 node cluster HA working with idrac/IPMI fencing. It is working when some node has a problem.
At this moment, for testing, 3 nodes are connected to one switch and I'm testing what happens if all is working and suddenly switch is powered off.
This is the state of the cluster before removing power to the switch:
Code:
clustatCluster Status for hcuscluster @ Wed May 27 07:29:24 2015
Member Status: Quorate
Member Name ID Status
------ ---- ---- ------
node1 1 Online, Local, rgmanager
node2 2 Online, rgmanager
node3 3 Online, rgmanager
Service Name Owner (Last) State
------- ---- ----- ------ -----
pvevm:100 node2 startedPVEVM has 192.168.1.144 and ping it's working:
Code:
C:\Users\>ping 192.168.1.144
Haciendo ping a 192.168.1.144 con 32 bytes de datos:
Respuesta desde 192.168.1.144: bytes=32 tiempo<1m TTL=64
Respuesta desde 192.168.1.144: bytes=32 tiempo<1m TTL=64
Respuesta desde 192.168.1.144: bytes=32 tiempo<1m TTL=64
Respuesta desde 192.168.1.144: bytes=32 tiempo<1m TTL=64At this point I cut the power to the switch and wait for some minute and power on again the switch.
The new info on clustat is this:
Code:
clustatTimed out waiting for a response from Resource Group Manager
Cluster Status for hcuscluster @ Wed May 27 07:49:44 2015
Member Status: Quorate
Member Name ID Status
------ ---- ---- ------
node1 1 Online, Local
node2 2 Online
node3 3 OnlineAnd pvevm 100 not starts again,ping is not possible host because is not accesible. I can't start manually, too. Is not possible that VM goes online when the switch is power on again? .
At this point is not possible to restart cman:
Code:
# /etc/init.d/cman restartStopping cluster:
Leaving fence domain... found dlm lockspace /sys/kernel/dlm/rgmanager
fence_tool: cannot leave due to active systems
[FAILED]Not possible restart rgmanager, it hangs on:
Code:
~# /etc/init.d/rgmanager restart
Stopping Cluster Service Manager:Is necessary to restart all nodes to get the inicial config and pvevm running?
Please, someone can help me How I can restart all services withouth restart nodes and how to pvevm goes online automated when switch is turn on again?
Thank you all.