On 07/07/2016 04:39 PM, Fran Barrera wrote:
Yes, this is the problem.
Well, you lose quorum once you stop A.
As the docs clearly state, you cannot tolerate failures if you have just
two monitors.
If your cluster only has two monitors, you cannot form quorum with just
one monitor: you need a majority up, running and able to communicate
among themselves.
Simply put, you need at least (n+1)/2 monitors up for a quorum to be
formed, 'n' being the total number of monitors in the cluster (i.e., in
the monmap).
You either need A and B to be running to be able to use the quorum, or
you need to add another monitor (call it C) so that you can stop A and
still have the cluster working.
-Joao
2016-07-07 17:34 GMT+02:00 Joao Eduardo Luis <joao@xxxxxxx
<mailto:joao@xxxxxxx>>:
On 07/07/2016 04:31 PM, Fran Barrera wrote:
Hello,
Yes I've added two monitors but the error persist. In the error
I see
only the IP of the first mon, why not appears the second?
The description you offered on the initial email appears to state
the following:
- You initially had one monitor (let's call it A)
- You added a second monitor (let's call it B)
- Everything works while A and B are running
- Nothing works if you stop A
Did I understand your problem correctly?
-Joao
I had only one monitors before and running good because I have
installed
AIO.
Thanks.
2016-07-07 17:22 GMT+02:00 Joao Eduardo Luis <joao@xxxxxxx
<mailto:joao@xxxxxxx>
<mailto:joao@xxxxxxx <mailto:joao@xxxxxxx>>>:
On 07/07/2016 04:17 PM, Fran Barrera wrote:
Hi all,
I have a cluster setup AIO with only one monitor and
now I've
created
another monitor in other server following this doc
http://docs.ceph.com/docs/master/rados/operations/add-or-rm-mons/ but
my
problem is if I stop the AIO monitor, the cluster stop
working.
It seems
like the ceph is not updated with the new mon or something
In the doc you quoted, one can read:
"Due to the nature of Paxos, Ceph requires a majority of
monitors
running to establish a quorum (thus establishing consensus).
[...]
For instance, on a 2 monitor deployment, no failures can be
tolerated in order to maintain a quorum; with 3 monitors, one
failure can be tolerated; [...]"
And in a box beneath, you also see
"Note: A majority of monitors in your cluster must be able
to reach
each other in order to establish a quorum."
So, say you have 2 monitors and you need a majority of them
to be
up, running, and able to communicate with each other in
order to
form quorum. What's a majority of 2? How many failures can
you tolerate?
-Joao
_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx <mailto:ceph-users@xxxxxxxxxxxxxx>
<mailto:ceph-users@xxxxxxxxxxxxxx
<mailto:ceph-users@xxxxxxxxxxxxxx>>
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
