2024-06-07T00:04:49.503Z

<Xiubo Li> Oh, cool. I just missed that PR.

2024-06-07T01:29:16.127Z

<Xiubo Li> I added one comments on @Venky Shankar's PR: <https://github.com/ceph/ceph/pull/57354#issuecomment-2153690011>

Just a little confuse about that PR, currently it will treat the `scrub` as pass when any of the `parent` is in replica, but what if there really has divergent between other parents ?

2024-06-07T11:41:07.793Z

<Mark Nelson (nhm)> Heya folks, is any of the data in the metadata pool not stored in OMAP?

2024-06-07T11:51:00.489Z

<Venky Shankar> mds journal is in the metadata pool.

2024-06-07T12:30:02.806Z

<Mark Nelson (nhm)> Ok, so for a mixed HDD/Flash cluster where the DB (ie OMAP) will be on flash, we still might want the meta pool on flash backed OSDs?

2024-06-07T12:40:31.012Z

<Venky Shankar> Yes, the journal are frequently written to with events and when the events are flushed out the metadata mutations are written to directory omap followed by trimming the journal.

2024-06-07T12:41:04.843Z

<Venky Shankar> Yes, the journal are frequently written to with events and when the events are flushed out, the metadata mutations are written to the directory omap followed by trimming the journal.

2024-06-07T16:38:05.207Z

<Patrick Donnelly> done