[MXS-1453] MaxScale consider healthy a server with full disk Created: 2017-09-26 Updated: 2020-08-25 Resolved: 2018-07-05 |
|
| Status: | Closed |
| Project: | MariaDB MaxScale |
| Component/s: | mariadbmon |
| Affects Version/s: | 2.0.0 |
| Fix Version/s: | 2.3.0 |
| Type: | Bug | Priority: | Major |
| Reporter: | Claudio Nanni | Assignee: | Johan Wikman |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | None | ||
| Environment: |
RHEL |
||
| Issue Links: |
|
||||||||||||||||
| Description |
|
If a backend server has the storage full MaxScale does not realize it and keeps sending traffic to the node. |
| Comments |
| Comment by markus makela [ 2017-11-29 ] |
|
Stopping the server with the full disk allows MaxScale to detect it. |
| Comment by Geoff Montee (Inactive) [ 2018-01-16 ] |
|
For GaleraMon, the best solution to this would probably be to change Galera, so that it terminates a node if the node's disk fills up. There is a feature request about this: https://github.com/codership/mysql-wsrep/issues/281 It wouldn't be enough for MaxScale to stop routing queries to the node with the full disk. Since all transactions have to be certified by all nodes, the node will continue causing problems for the cluster until it is removed from the cluster. Therefore, the best solution would probably be for Galera to auitomatically remove the node from the cluster. |
| Comment by Johan Wikman [ 2018-07-05 ] |
|
Fixed by |