On Thu, 14 Aug 2014, Pawe? Sadowski wrote: > W dniu 14.08.2014 17:20, Sage Weil pisze: > > On Thu, 14 Aug 2014, Pawe? Sadowski wrote: > >> Hello, > >> > >> I've a cluster of 35 OSD (30 HDD, 5 SSD) with cache tiering configured. > >> During tests it looks like ceph is not respecting target_max_bytes > >> settings. Steps to reproduce: > >> - configure cache tiering > >> - set target_max_bytes to 32G (on hot pool) > >> - write more than 32G of data > >> - nothing happens <snip details> The reason the agent is doing work is because you don't have hit_set_* configured for the cache pool, which means the cluster isn't tracking what objects get read to inform the flush/evict decisions. Configuring that will fix this. Try ceph osd pool set cache hit_set_type bloom ceph osd pool set cache hit_set_count 8 ceph osd pool set cache hit_set_period 3600 or similar. The agent could still run in a brain-dead mode without it, but it suffers from the bug you found. That was fixed after 0.80.5 and will be in 0.80.6. Thanks! sage
 |