[Maria-discuss] Replication-manager + Maxscale
Hi Experts, I have a 2 db server with one as master and another one as slave. Maxscale implemented with below monitor config: *[MySQL Monitor]* *type=monitor* *module=mysqlmon* *servers=master1,slave1* *monitor_interval=500* *script=/usr/bin/replication-manager --user root:password --rpluser replusr:password * *--hosts $INITIATOR,$NODELIST --failover=force --interactive=false* *events=master_down* When I shutdown master, failover is not automated. MRM (Replication Manager) is not promoting the slave to master automatically. In maxscale log, I noticed below: *2017-03-02 03:33:36 notice : Executed monitor script '/usr/bin/replication-manager --user root:password --rpluser replusr:password --hosts $INITIATOR,$NODELIST --failover=force --interactive=false' on event 'master_down'.* *2017-03-02 03:33:36 error : No Master can be determined.* Is this because I didn't have the quorum (no slave) since its one master and one slave set-up. Is there any workaround. Regards, Kart
Hi Karthick, (disclaimer: I am one of the authors of Replication Manager for MariaDB) you seem to be using outdated instructions. Please grab the 1.0.2 release for your distribution there: https://github.com/tanji/replication-manager/releases Replication Manager works best when used independently from MaxScale. Please read our documentation at https://github.com/tanji/replication-manager/blob/develop/README.md and let me know if you have any questions. Best Guillaume Le ven. 3 mars 2017 à 00:05, Karthick Subramanian < ksubramanian@paycommerce.com> a écrit :
Hi Experts,
I have a 2 db server with one as master and another one as slave.
Maxscale implemented with below monitor config:
*[MySQL Monitor]* *type=monitor* *module=mysqlmon* *servers=master1,slave1* *monitor_interval=500* *script=/usr/bin/replication-manager --user root:password --rpluser replusr:password * *--hosts $INITIATOR,$NODELIST --failover=force --interactive=false* *events=master_down*
When I shutdown master, failover is not automated. MRM (Replication Manager) is not promoting the slave to master automatically.
In maxscale log, I noticed below:
*2017-03-02 03:33:36 notice : Executed monitor script '/usr/bin/replication-manager --user root:password --rpluser replusr:password --hosts $INITIATOR,$NODELIST --failover=force --interactive=false' on event 'master_down'.* *2017-03-02 03:33:36 error : No Master can be determined.*
Is this because I didn't have the quorum (no slave) since its one master and one slave set-up. Is there any workaround.
Regards, Kart
_______________________________________________ Mailing list: https://launchpad.net/~maria-discuss Post to : maria-discuss@lists.launchpad.net Unsubscribe : https://launchpad.net/~maria-discuss More help : https://help.launchpad.net/ListHelp
Hi Karthick, Please also use the upcoming release 1.1 with is the current dev release on github This is the best way to make this setup running I ‘ll send you a package if you tell me for what os ? /stephane Stéphane Varoqui, Senior Consultant Phone: +33 695-926-401, skype: svaroqui http://www.mariadb.com <http://www.mariadb.com/>
Le 2 mars 2017 à 17:26, Guillaume Lefranc <guillaume@adishatz.net> a écrit :
Hi Karthick,
(disclaimer: I am one of the authors of Replication Manager for MariaDB)
you seem to be using outdated instructions. Please grab the 1.0.2 release for your distribution there: https://github.com/tanji/replication-manager/releases <https://github.com/tanji/replication-manager/releases>
Replication Manager works best when used independently from MaxScale. Please read our documentation at https://github.com/tanji/replication-manager/blob/develop/README.md <https://github.com/tanji/replication-manager/blob/develop/README.md> and let me know if you have any questions.
Best Guillaume
Le ven. 3 mars 2017 à 00:05, Karthick Subramanian <ksubramanian@paycommerce.com <mailto:ksubramanian@paycommerce.com>> a écrit : Hi Experts,
I have a 2 db server with one as master and another one as slave.
Maxscale implemented with below monitor config:
[MySQL Monitor] type=monitor module=mysqlmon servers=master1,slave1 monitor_interval=500 script=/usr/bin/replication-manager --user root:password --rpluser replusr:password --hosts $INITIATOR,$NODELIST --failover=force --interactive=false events=master_down
When I shutdown master, failover is not automated. MRM (Replication Manager) is not promoting the slave to master automatically.
In maxscale log, I noticed below:
2017-03-02 03:33:36 notice : Executed monitor script '/usr/bin/replication-manager --user root:password --rpluser replusr:password --hosts $INITIATOR,$NODELIST --failover=force --interactive=false' on event 'master_down'. 2017-03-02 03:33:36 error : No Master can be determined.
Is this because I didn't have the quorum (no slave) since its one master and one slave set-up. Is there any workaround.
Regards, Kart
_______________________________________________ Mailing list: https://launchpad.net/~maria-discuss <https://launchpad.net/~maria-discuss> Post to : maria-discuss@lists.launchpad.net <mailto:maria-discuss@lists.launchpad.net> Unsubscribe : https://launchpad.net/~maria-discuss <https://launchpad.net/~maria-discuss> More help : https://help.launchpad.net/ListHelp <https://help.launchpad.net/ListHelp> _______________________________________________ Mailing list: https://launchpad.net/~maria-discuss Post to : maria-discuss@lists.launchpad.net Unsubscribe : https://launchpad.net/~maria-discuss More help : https://help.launchpad.net/ListHelp
Hi Stephane, Once I have tested my settings, I am planning to use it in production version. That's why I thought of using GA release( https://github.com/tanji/replication-manager/releases/tag/1.0.0). But please send me for the below details, So I will test my settings: OS : CentOS 7 Maria DB : 10.1.17 Maxscale : 2:0 Hi Guillaume, Thank you for your input. In order to make MRM independent from Maxscale, I don't need to configure to execute as script via mysql Monitor section of maxscale for event=master_down? Also can you share me some documentation to understand this better, I am little bit new to this technology and facing little challenge to understand these below: Also I am planning to use pcs-corosync to make maxscale(Active-standby) HA. Source: https://github.com/tanji/replication-manager/blob/develop/README.md
- With monitor-less proxies, replication-manager can call scripts that set and reload the new configuration of the leader route. A common scenario is an VRRP Active Passive HAProxy sharing configuration via a network disk with thereplication-manager scripts
- Using replication-manager as an API component of a group communication cluster. MRM can be called as a Pacemaker resource that moves alongside a VIP, the monitoring of the cluster is in this case already in charge of the GCC.
On Thu, Mar 2, 2017 at 10:02 PM, Stephane Varoqui <stephane@mariadb.com> wrote:
Hi Karthick,
Please also use the upcoming release 1.1 with is the current dev release on github This is the best way to make this setup running I ‘ll send you a package if you tell me for what os ?
/stephane
Stéphane Varoqui, Senior Consultant Phone: +33 695-926-401, skype: svaroqui http://www.mariadb.com
Le 2 mars 2017 à 17:26, Guillaume Lefranc <guillaume@adishatz.net> a écrit :
Hi Karthick,
(disclaimer: I am one of the authors of Replication Manager for MariaDB)
you seem to be using outdated instructions. Please grab the 1.0.2 release for your distribution there: https://github.com/tanji/replication-manager/releases
Replication Manager works best when used independently from MaxScale. Please read our documentation at https://github.com/tanji/ replication-manager/blob/develop/README.md and let me know if you have any questions.
Best Guillaume
Le ven. 3 mars 2017 à 00:05, Karthick Subramanian < ksubramanian@paycommerce.com> a écrit :
Hi Experts,
I have a 2 db server with one as master and another one as slave.
Maxscale implemented with below monitor config:
*[MySQL Monitor]* *type=monitor* *module=mysqlmon* *servers=master1,slave1* *monitor_interval=500* *script=/usr/bin/replication-manager --user root:password --rpluser replusr:password * *--hosts $INITIATOR,$NODELIST --failover=force --interactive=false* *events=master_down*
When I shutdown master, failover is not automated. MRM (Replication Manager) is not promoting the slave to master automatically.
In maxscale log, I noticed below:
*2017-03-02 03:33:36 notice : Executed monitor script '/usr/bin/replication-manager --user root:password --rpluser replusr:password --hosts $INITIATOR,$NODELIST --failover=force --interactive=false' on event 'master_down'.* *2017-03-02 03:33:36 error : No Master can be determined.*
Is this because I didn't have the quorum (no slave) since its one master and one slave set-up. Is there any workaround.
Regards, Kart
_______________________________________________ Mailing list: https://launchpad.net/~maria-discuss Post to : maria-discuss@lists.launchpad.net Unsubscribe : https://launchpad.net/~maria-discuss More help : https://help.launchpad.net/ListHelp
_______________________________________________ Mailing list: https://launchpad.net/~maria-discuss Post to : maria-discuss@lists.launchpad.net Unsubscribe : https://launchpad.net/~maria-discuss More help : https://help.launchpad.net/ListHelp
Le 2 mars 2017 à 17:45, Karthick Subramanian <ksubramanian@paycommerce.com> a écrit :
Hi Stephane,
Once I have tested my settings, I am planning to use it in production version. That's why I thought of using GA release(https://github.com/tanji/replication-manager/releases/tag/1.0.0 <https://github.com/tanji/replication-manager/releases/tag/1.0.0>).
But please send me for the below details, So I will test my settings:
OS : CentOS 7 Maria DB : 10.1.17 Maxscale : 2:0
Hi Guillaume,
Thank you for your input. In order to make MRM independent from Maxscale, I don't need to configure to execute as script via mysql Monitor section of maxscale for event=master_down?
Yes that is correct , just run replication-manager in monitoring mode and daemon. The rpm should come with configuration file that you need to adapt in /etc/replication-manger and rename to config.toml so that that starting the service work No need to instruct MaxScale at all In release 1.0 you only have the choice to let maxscale detect about the new topology and that will work for switchover but no failover in release 1.1 you can just tell replication-manager to drive maxscale so that failover works as well Hope it helps !
Also can you share me some documentation to understand this better, I am little bit new to this technology and facing little challenge to understand these below:
Also I am planning to use pcs-corosync to make maxscale(Active-standby) HA.
Source: https://github.com/tanji/replication-manager/blob/develop/README.md <https://github.com/tanji/replication-manager/blob/develop/README.md>
With monitor-less proxies, replication-manager can call scripts that set and reload the new configuration of the leader route. A common scenario is an VRRP Active Passive HAProxy sharing configuration via a network disk with thereplication-manager scripts Using replication-manager as an API component of a group communication cluster. MRM can be called as a Pacemaker resource that moves alongside a VIP, the monitoring of the cluster is in this case already in charge of the GCC.
On Thu, Mar 2, 2017 at 10:02 PM, Stephane Varoqui <stephane@mariadb.com <mailto:stephane@mariadb.com>> wrote: Hi Karthick,
Please also use the upcoming release 1.1 with is the current dev release on github This is the best way to make this setup running I ‘ll send you a package if you tell me for what os ?
/stephane Stéphane Varoqui, Senior Consultant Phone: +33 695-926-401, skype: svaroqui http://www.mariadb.com <http://www.mariadb.com/>
Le 2 mars 2017 à 17:26, Guillaume Lefranc <guillaume@adishatz.net <mailto:guillaume@adishatz.net>> a écrit :
Hi Karthick,
(disclaimer: I am one of the authors of Replication Manager for MariaDB)
you seem to be using outdated instructions. Please grab the 1.0.2 release for your distribution there: https://github.com/tanji/replication-manager/releases <https://github.com/tanji/replication-manager/releases>
Replication Manager works best when used independently from MaxScale. Please read our documentation at https://github.com/tanji/replication-manager/blob/develop/README.md <https://github.com/tanji/replication-manager/blob/develop/README.md> and let me know if you have any questions.
Best Guillaume
Le ven. 3 mars 2017 à 00:05, Karthick Subramanian <ksubramanian@paycommerce.com <mailto:ksubramanian@paycommerce.com>> a écrit : Hi Experts,
I have a 2 db server with one as master and another one as slave.
Maxscale implemented with below monitor config:
[MySQL Monitor] type=monitor module=mysqlmon servers=master1,slave1 monitor_interval=500 script=/usr/bin/replication-manager --user root:password --rpluser replusr:password --hosts $INITIATOR,$NODELIST --failover=force --interactive=false events=master_down
When I shutdown master, failover is not automated. MRM (Replication Manager) is not promoting the slave to master automatically.
In maxscale log, I noticed below:
2017-03-02 03:33:36 notice : Executed monitor script '/usr/bin/replication-manager --user root:password --rpluser replusr:password --hosts $INITIATOR,$NODELIST --failover=force --interactive=false' on event 'master_down'. 2017-03-02 03:33:36 error : No Master can be determined.
Is this because I didn't have the quorum (no slave) since its one master and one slave set-up. Is there any workaround.
Regards, Kart
_______________________________________________ Mailing list: https://launchpad.net/~maria-discuss <https://launchpad.net/~maria-discuss> Post to : maria-discuss@lists.launchpad.net <mailto:maria-discuss@lists.launchpad.net> Unsubscribe : https://launchpad.net/~maria-discuss <https://launchpad.net/~maria-discuss> More help : https://help.launchpad.net/ListHelp <https://help.launchpad.net/ListHelp> _______________________________________________ Mailing list: https://launchpad.net/~maria-discuss <https://launchpad.net/~maria-discuss> Post to : maria-discuss@lists.launchpad.net <mailto:maria-discuss@lists.launchpad.net> Unsubscribe : https://launchpad.net/~maria-discuss <https://launchpad.net/~maria-discuss> More help : https://help.launchpad.net/ListHelp <https://help.launchpad.net/ListHelp>
participants (3)
-
Guillaume Lefranc -
Karthick Subramanian -
Stephane Varoqui