Description:
Use case / background:
A customer had the following idea which I thought I should suggest as a new feature and might be at least reasonably simple to implement. (This is a convenience request, not critical.)
The general context is a site running multiple MySQL clusters. These could be independent databases or replicating each other. Further, log messages are consolidated by a 3rd party or proprietary monitoring server. The problem in such a setting is that a single log line doesn't contain any hostname or other info where it would be immediately obvious *which* cluster the message is from.
Ex.
2007-12-20 08:20:43 [MgmSrvr] ALERT -- Node 1: Node 3 Disconnected
We know that Node 3 disconnected, but in which cluster?
The same question is also valid in other use cases where log messages are detached from their original context, for instance copy-pasted into an email.
It is worth mentioning that currently the monitoring application/server could insert Cluster identification information itself, based on the source the messaging arrived from (each cluster has separate log file, etc...). Yet, it would be convenient if it is possible to implement directly in the MySQL Cluster.
How to repeat:
tail /var/log/mysql.log
Suggested fix:
For instance:
Introduce parameter
[ndb default]
ClusterName= Ebba
And output logs as
2007-12-20 08:20:43 [Ebba MgmSrvr] ALERT -- Node 1: Node 3 Disconnected
Alternatively, use current HostName parameter and add that into log. Yet this is often an ip address, thus not as informative and is also longer than a name might be. (And it is per node name, not per cluster.)
© 2024, Oracle Corporation and/or its affiliates