Expired AlarmChanges doesn't delete by ceilometer-expirer

[Mitsuru Kanabuchi]

Hi,

Currently the ceilometer-expirer doesn't delete expired AlarmChanges.
Remained AlarmChanges would be cause of wasted disk-space and slow response.

I think, the ceilometer-expirer should delete expired AlarmChanges with other expired datas.

Cheers,

[gordon chung (chungg)]

hmm. do you believe there should be two separate processes for this? seems like alarm expirations shouldn't be the same frequency as sample expiration.

[Mitsuru Kanabuchi (kanabuchi)]

Hi Gordon, thank you for your reply.

i agree that alarm expirations shouldn't be the same frequency as sample expiration.
i think we need to separate TTL(CONF.database.time_to_live), for alarm expiration and sample expiration.
we can separate processes as well.

[Mitsuru Kanabuchi (kanabuchi)]

We are planning to fix it as follows:

1. Delete expired AlarmChanges by separate process (other than ceilometer-expirer)
2. Add another TTL (other than the CONF.database.time_to_live)

Let us know if you have some comments.

[Openstack Gerrit (openstack-gerrit)]

Fix proposed to branch: master
Review: https://review.openstack.org/87869

[gordon chung (chungg)]

fixed in above patch