Fluentd flushes to elasticsearch with an impatient delay.
Categorize the logs in their various indices.Write all the configs and include them in thecollector.conf- Set up log rotation. Use curator
- Rabbitmq and Mysql log config
- Parse the request tokens into request id, user id, and tenant id. Look here.
- Parse the log file
/var/log/apache2/keystone.log Parse the nova-api.log file properly. Skip the stack trace.
-
Fluentd
- Log level: TRACE
- IP: 0.0.0.0 (Default is 0.0.0.0)
- Port: 24220 (Default is 24224)
-
Elasticsearch
- Clustername: sdcloud
- IP: 0.0.0.0 (Default is 0.0.0.0)
- Port: 9200 (Default is 9200)
- Flush interval: 1s
- Index name: Second tag part
- Type name: Third tag part
-
Connet
- Forward to
- Active: 10.41.0.112:24220
- Standby: 10.41.0.111:24220
- Forward to
-
Compute
- Forward to
- Active: 10.41.0.112:24220
- Standby: 10.41.0.111:24220
- Forward to