hcq/MaxScale
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Clarify documentation on verify_master_failure

Browse Source
This commit is contained in:
Esa Korhonen
2018-01-23 15:45:42 +02:00
parent d681d0f2fa
commit c86ee1f53a
1 changed files with 14 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

38
Documentation/Monitors/MariaDB-Monitor.md
View File

@ -379,34 +379,24 @@ from passive to active will wait for a failover to take place after an apparent
loss of a master server. If no new master server is detected within the loss of a master server. If no new master server is detected within the
configured time period, failover will be initiated again. configured time period, failover will be initiated again.
#### `verify_master_failure` #### `verify_master_failure` and `master_failure_timeout`
Enable master failure verification for automatic failover. This parameter Enable additional master failure verification for automatic failover.
expects a boolean value and the feature is enabled by default. `verify_master_failure` is a boolean value (default: true) which enables this
feature and `master_failure_timeout` defines the timeout in seconds (default: 10).
The failure of a master can be verified by checking whether the slaves are still The failure verification is performed by checking whether the slaves are still
connected to the master. The timeout for master failure verification is connected to the master and receiving events. Effectively, if a slave has
controlled by the `master_failure_timeout` parameter. received an event within `master_failure_timeout` seconds, the master is not
considered down when deciding whether to auto_failover.
#### `master_failure_timeout` If every slave loses its connection to the master (*Slave_IO_Running* is not
"Yes"), master failure is considered verified regardless of timeout. This allows
a faster failover when the master server crashes, as that causes immediate
disconnection.
This parameter controls the period of time, in seconds, that the monitor must For automatic failover to activate, the `failcount` requirement must also be
wait before it can declare that the master has failed. The default value is 10 met.
seconds. For failover to activate, the `failcount` requirement must also be met.
The failure of a master is verified by tracking when the last change to the
relay log was done and when the last replication heartbeat was received. If the
period of time between the last received event and the time of the check exceeds
the configured value, the slave's connection to the master is considered to be
broken.
When all slaves of a failed master are no longer connected to the master, the
master failure is verified and the failover can be safely performed.
If the slaves lose their connections to the master before the configured timeout
is exceeded, the failover is performed immediately. This allows a faster
failover when the master server crashes causing immediate disconnection of the
the network connections.
#### `switchover_timeout` #### `switchover_timeout`