Update to v25.02
Update procedure for v25.02 does not differ from the standard one, but requires additional steps:
- metrics migration;
- (optional) settings for distributing requests for plotting graphs;
- (optional) setting thresholds for incident registration.
Metrics Migration Procedure
-
Download the migration file to all ClickHouse instances.
wget https://docs.mitigator.ru/master/kb/update2502/update2502.sql -O /srv/mitigator/update2502.sql
-
Run the migration:
docker-compose exec -T clickhouse clickhouse-client -mn < update2502.sql
No messages are printed in case of success. Please report any error messages to the support.
-
Restart Carbon Clickhouse:
docker-compose restart carbon-clickhouse
Migration may take a while. DDoS-protection functions of MITIGATOR are fully operational while migrating data. Graphs for time ranges in the past may not render during migration (current metrics will be rendered).
Procedure for Setting Up Distribution of Requests for Plotting Graphs
In version v25.02, the ability to evenly distribute graph requests across metric
storages has been added.
The list of servers to read is specified by the CARBONAPI_BACKENDS
variable
of the carbonapi
service.
For the default scheme (two nodes with storages), it is enough to
-
update the
docker-compose.failover.yml
file on the nodes with the storage:wget https://docs.mitigator.ru/master/dist/multi/docker-compose.failover.yml
-
update the
docker-compose.worker.failover.yml
file on nodes without storage:wget https://docs.mitigator.ru/master/dist/multi/docker-compose.worker.failover.yml
For schemes with three or more failover storages, you need to add a section with the
carbonapi
service to the docker-compose.failover.yml
and docker-compose.worker.failover.yml
files and list the addresses of all nodes with the storage in the CARBONAPI_BACKENDS
variable:
carbonapi:
environment:
CARBONAPI_BACKENDS: "http://${SERVER1}:9090,http://${SERVER2}:9090,http://${SERVER3}:9090"
Autodetection Thresholds for Incident Registration
When upgrading to v25.02, Incidents.
thresholds with zero values will be removed.
Incidents in protection policies without Incidents.
thresholds will not be registered.
It is recommended to check the thresholds for registering incidents in protection policies
and set the required values where necessary.