On 09/01/2013 03:35 AM, Kasper Dieter wrote:
Hi, under http://eu.ceph.com/docs/wip-rpm-doc/config-cluster/rbd-config-ref/ I found a good description about RBD cache parameters.
You're looking at an old branch there - the current description is a bit more clear that this doesn't affect rbd.ko at all:
http://eu.ceph.com/docs/master/rbd/rbd-config-ref/
But, I am missing information - by whom these parameters are evaluated and - when will this happen ?
They will be read by librbd when an image is opened. For qemu, this means detaching and reattaching a disk will re-read the settings. They can't be changed on the fly right now, though it's possible to make that work with librbd itself. Note that qemu needs to think it's doing writeback caching in order to pass guest flushes through to librbd. Josh
My assumption: - the rbd_cache* parameter will be read by MONs - rbd.ko will contact MON during 'insmod' or during 'open /dev/rbd*' and get the rbd_cache info Please confirm / correct. Can I change the parameter only via ceph.conf + restart of MONs, or can I change rbd_cache on runtime? e.g. # ceph --admin-daemon /var/run/ceph/ceph-mon.0.asok config show | grep rbd_cache "rbd_cache": "false", "rbd_cache_writethrough_until_flush": "false", "rbd_cache_size": "33554432", "rbd_cache_max_dirty": "25165824", "rbd_cache_target_dirty": "16777216", "rbd_cache_max_dirty_age": "1", "rbd_cache_block_writes_upfront": "false", # ceph mon tell 0 injectargs '--rbd_cache true' # ceph --admin-daemon /var/run/ceph/ceph-mon.0.asok config show | grep rbd_cache "rbd_cache": "true", "rbd_cache_writethrough_until_flush": "false", "rbd_cache_size": "33554432", "rbd_cache_max_dirty": "25165824", "rbd_cache_target_dirty": "16777216", "rbd_cache_max_dirty_age": "1", "rbd_cache_block_writes_upfront": "false", Regards, -Dieter _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
_______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
