If you have igmp snooping (filtering) somewhere, you can have the case of omping works, and corosync not working.
(because both use differents multicast address, I don't remember, but omping have a default specific multicast address, and corosync create an address from the cluster name)
So your cisco switchs (with IGMP snooping : Enabled), create multicast groups by multicast address, each groups have physical ports. (sh igmp snooping groups)
So you can have a working group for omping, and a non working group for corosync.
So, first thing to check if is corosync works.
# /etc/init.d/cman restart
# pvecm status
check on your cisco switchs that group is ok (sh igmp snooping groups)
If that's ok, then you can restart
#/etc/iinit.d/pve-cluster restart -> so, you can write in /etc/pve/.....
Then finally restart
#/etc/init.d/pvestatd -> your node should come back in green
But try to disable igmp snooping for your switchs first to see if it's helping. (you can enable/disable it by vlan I think)
I don't know for your cisco switchs, by mine, show ip igmp snooping groups
your can notice differents mutlicast groups with differents ports
Edit:
default multicast address:
omping: 232.43.211.234
corosync:
By default it will generate a multicast address using 239.192.x.x where x.x is the 16bit cluster ID number split into bytes.
you can see the multicast address of corosync with "pvecm status"
So for test, you need to find use omping with your corosync multicast address
"omping -m 239.192.x.x ...."
(because both use differents multicast address, I don't remember, but omping have a default specific multicast address, and corosync create an address from the cluster name)
So your cisco switchs (with IGMP snooping : Enabled), create multicast groups by multicast address, each groups have physical ports. (sh igmp snooping groups)
So you can have a working group for omping, and a non working group for corosync.
So, first thing to check if is corosync works.
# /etc/init.d/cman restart
# pvecm status
check on your cisco switchs that group is ok (sh igmp snooping groups)
If that's ok, then you can restart
#/etc/iinit.d/pve-cluster restart -> so, you can write in /etc/pve/.....
Then finally restart
#/etc/init.d/pvestatd -> your node should come back in green
But try to disable igmp snooping for your switchs first to see if it's helping. (you can enable/disable it by vlan I think)
I don't know for your cisco switchs, by mine, show ip igmp snooping groups
your can notice differents mutlicast groups with differents ports
Code:
# show ip igmp snooping groups
Vlan Group Type Version Port List
-----------------------------------------------------------------------
94 239.192.3.235 igmp v2 Gi0/3, Gi0/9,
Gi0/24
94 239.192.13.22 igmp v2 Gi0/5, Gi0/7,
Gi0/10, Gi0/12,
Gi0/24
sw-dist33# show ip igmp snooping groups vlan 94
Vlan Group Type Version Port List
-----------------------------------------------------------------------
94 239.192.3.235 igmp v2 Gi0/3, Gi0/9,
Gi0/24
94 239.192.13.22 igmp v2 Gi0/5, Gi0/7,
Gi0/10, Gi0/12,
Gi0/24
sw-dist33#Edit:
default multicast address:
omping: 232.43.211.234
corosync:
By default it will generate a multicast address using 239.192.x.x where x.x is the 16bit cluster ID number split into bytes.
you can see the multicast address of corosync with "pvecm status"
So for test, you need to find use omping with your corosync multicast address
"omping -m 239.192.x.x ...."
Last edited: