1) primary up
2) secondary up
3) both fully synchronized
4) clients writes to the cluster
5) in this moment primary crashes (switched off for simulation)
6) former secondary takes over the service and is now primary, the other node is unknown
7) now a power failure happens, the new primary goes down smoothly
8) after power is back, the crashed node is still crashed
9) the only working node, former secondary, now primary, boots
10) it asks, if it should become primary
11) I answer yes
12) NOTHING happens, I even cannot switch to virtual terminal 2
13) the only thing which helps is to boot the formerly crashed node too
14) then the primary is able to become primary
Why is this like described? Shouldn't it start in stand alone mode?
With this behavior of DRBD, there MUST be no reboot of a node if the other is crashed and is to
be repaired, i.e. new disk. So if I have to wait three days for a new disk, I have no high
availability meanwhile, this is ok. But with a simple reboot of the remaining node, I have NO
availability at all.
I've asked this already, but there was no reaction.
mfg ar
--
mailto:andreas@example.com
http://www.rittershofer.de
PGP-Public-Key http://www.rittershofer.de/ari.htm
2) secondary up
3) both fully synchronized
4) clients writes to the cluster
5) in this moment primary crashes (switched off for simulation)
6) former secondary takes over the service and is now primary, the other node is unknown
7) now a power failure happens, the new primary goes down smoothly
8) after power is back, the crashed node is still crashed
9) the only working node, former secondary, now primary, boots
10) it asks, if it should become primary
11) I answer yes
12) NOTHING happens, I even cannot switch to virtual terminal 2
13) the only thing which helps is to boot the formerly crashed node too
14) then the primary is able to become primary
Why is this like described? Shouldn't it start in stand alone mode?
With this behavior of DRBD, there MUST be no reboot of a node if the other is crashed and is to
be repaired, i.e. new disk. So if I have to wait three days for a new disk, I have no high
availability meanwhile, this is ok. But with a simple reboot of the remaining node, I have NO
availability at all.
I've asked this already, but there was no reaction.
mfg ar
--
mailto:andreas@example.com
http://www.rittershofer.de
PGP-Public-Key http://www.rittershofer.de/ari.htm