[Maria-discuss] Problem when all galera nodes are down
Hi again, We're running 3 galera mariadb 5.5 servers on Ubuntu 12.04 AMD 64 server. With one drbd server on another host. We have a specific case in our dev environment. All the hosts are stopped at night. So when they all boot up at the same time in the morning, each server is looking for another one up. Since there are no other server up, they refuse to start. I have to edit the wsrep_cluster_address variable on one server to set it back to "gcomm://" and start it. Then start the 2 others servers and once everything is up, put back the wsrep_cluster_address variable to its previous value and restart the server. Galera isn't prepared having all its nodes down ? Thanks, -- Guillaume Hilt
On 2013-10-29 10:48, Guillaume Hilt wrote:
Hi again,
We're running 3 galera mariadb 5.5 servers on Ubuntu 12.04 AMD 64 server. With one drbd server on another host.
We have a specific case in our dev environment. All the hosts are stopped at night.
So when they all boot up at the same time in the morning, each server is looking for another one up. Since there are no other server up, they refuse to start.
I have to edit the wsrep_cluster_address variable on one server to set it back to "gcomm://" and start it. Then start the 2 others servers and once everything is up, put back the wsrep_cluster_address variable to its previous value and restart the server.
Galera isn't prepared having all its nodes down ?
As you can see. Not at the moment.
Thanks,
-- Alexey Yurchenko, Codership Oy, www.codership.com Skype: alexey.yurchenko, Phone: +358-400-516-011
Ok, I thought it was a misconfiguration on my side :) Well, it's easy to restart anyway, so it's no big deal. Thanks. Guillaume Hilt Le 29/10/2013 13:02, Alex Yurchenko a écrit :
On 2013-10-29 10:48, Guillaume Hilt wrote:
Hi again,
We're running 3 galera mariadb 5.5 servers on Ubuntu 12.04 AMD 64 server. With one drbd server on another host.
We have a specific case in our dev environment. All the hosts are stopped at night.
So when they all boot up at the same time in the morning, each server is looking for another one up. Since there are no other server up, they refuse to start.
I have to edit the wsrep_cluster_address variable on one server to set it back to "gcomm://" and start it. Then start the 2 others servers and once everything is up, put back the wsrep_cluster_address variable to its previous value and restart the server.
Galera isn't prepared having all its nodes down ?
As you can see. Not at the moment.
Thanks,
participants (2)
-
Alex Yurchenko -
Guillaume Hilt