[MXS-3296] [Enhancement] Cooperative Monitoring for [Galera Monitor] or [Common Monitor] Created: 2020-11-13  Updated: 2022-12-03  Resolved: 2022-09-08

Status: Closed
Project: MariaDB MaxScale
Component/s: Monitor
Affects Version/s: None
Fix Version/s: N/A

Type: New Feature Priority: Major
Reporter: Kwangbock Lee Assignee: Todd Stoffel (Inactive)
Resolution: Won't Do Votes: 1
Labels: None


 Description   

Currently Cooperative Monitoring in MaxScale 2.5 only works in `MariaDB Monitor`.
It would be great if we can also add this feature to `Galera Monitor` or to common monitor parameters.

Best
Kwangbock



 Comments   
Comment by markus makela [ 2020-11-13 ]

Galera does not need this as it manages replication on its own. You can use root_node_as_master=true to force all MaxScales to use the same node as the master.

There is some use for this: being able to determine which MaxScale performs tasks that must only be done once. In 2.6 this will include the replication of data from MariaDB into external systems (avrorouter and kafkacdc) so there's some merit to having something to coordinate this with. For Galeramon, this would be as simple as acquiring a lock on the node currently assigned as the Master.

Comment by Avijit Batabyal [ 2021-12-08 ]

"For Galeramon, this would be as simple as acquiring a lock on the node currently assigned as the Master."
But there is no flag as "Primary" for galeramon to determine if a maxscale instance has acquired lock on master node.
like this

Monitor Diagnostics │ {                                                          │
│                     │     "master": "server1",                             │
│                     │     "master_gtid_domain_id": 0,                            │
│                     │     "primary": true,                                       │
│                     │     "server_info": [                                       │
│                     │         {                         

Comment by markus makela [ 2021-12-08 ]

Yes, this would have to be added if this was done for Galeramon as well.

Comment by Avijit Batabyal [ 2021-12-09 ]

ok. Actually we are planning use multiple maxscale instances monitoring the same galera backends. We can do it using Mariadb monitor for primary-replica setup.
Could you provide any tentative timeline for this feature to be implemented in available stable releases of maxscale ?

Comment by markus makela [ 2021-12-09 ]

At this point in time I don't think we can give an estimate.

Comment by Avijit Batabyal [ 2022-09-08 ]

any reason please, why this "won't be" implemented?

Comment by ssauravy [ 2022-11-24 ]

In a maxscale configuration of 2 or more units
Deadlock occurs when using galeramon.
A problem arises.

Comment by markus makela [ 2022-11-24 ]

kyoung-yeon are you using root_node_as_master=true in Galeramon?

Comment by ssauravy [ 2022-11-24 ]

What if the node with root_node_as_master=true and wsrep_local_index=0 goes down?

Comment by markus makela [ 2022-11-24 ]

The Galera cluster will reorganize itself and reassign the nodes so that another node will be the one with wsrep_local_index=0.

Generated at Thu Feb 08 04:20:23 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.