dfilppi / cloudify-healer-plugin Goto Github PK

Add ability to detect failure by failure to open TCP port

An HA implementation will require awareness of leader startup. After failover and leader election, the manager needs to provide a notification to deployments that have such long running services or they won't restart.

As a sanity check, periodically and before heal, check to make sure the related deployment still exists. If not, die.

Add the ability for the user to point at a user-defined script external to the plugin that detect failures

Currently the http healer just looks for a 2xx response to define success. In addition, it would be good to allow a regex (or something) to be specified to look for indicators in the response body.

If possible, try to eliminate need to pass in explicit rest api connection creds. Ideally use creds of user of deployment automagically.

Add ability to detect failure via http GET

Bad on a few levels. The healer needs to wait for install completion because:

1. may be configured to fail fast, and then it will attempt a heal, fail at it because install is running, and exit
2. on a failed installation, the healer node may be complete, and start trying to heal things that don't exist or shouldn't exist.