For now, just append this to the end of the pg <pgid> query json dump.
We definitely want to do something smarter here, but I'm not sure whether
json or plaintext is the way to go.
Signed-off-by: Sage Weil <sage@newdream.net>
Dump missing objects in json. If more key is non-zero, user should ask for
more by passing the last object as the offset for the next request.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
This is lightweight and relies on boost spirit, which we already use, so
there are no new dependencies.
There were some other libraries that also looked good, but they weren't
already packages for existing Debian distros like squeeze or even wheezy.
Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
This doesn't exist any more, and I don't think it
ever "cleanly shut down the filesystem" -- certainly not
within my recent lifetime!
Signed-off-by: Greg Farnum <gregory.farnum@dreamhost.com>
Reviewed-by: Dan Mick <dan.mick@dreamhost.com>
We need to hold the lock for ms_dispatch, esp calls into objecter. We
should only drop it when blocking; use distinct naming for the on-stack
mutex used for that.
Reported-by: Alexandre Oliva <oliva@lsd.ic.unicamp.br>
Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
If we have a temporary mapping for this PG, consider that unclean. This
makes CLEAN and REMAPPED mutually exclusive. For example, a 2 node cluster
with 2x replication and one osd marked out will make the pgs all
active+remapped, not active+clean+remapped.
Fixes: #2094
Signed-off-by: Sage Weil <sage@newdream.net>
This fixes OSDMap compatibility between v0.42 and <v0.42.
For MOSDMap, reencode maps if OSDENC feature is missing. Also rev the
message version. We don't use COMPAT version here because v3 can't be
understood by v2 (that's why we're checking feature bits). (It will be
possible to do that later when our constituent types can be decoded by
multiple versions.)
Fixes: #2095
Signed-off-by: Sage Weil <sage@newdream.net>
Reviewed-by: Greg Farnum <gregory.farnum@dreamhost.com>
Otherwise the MDS will leave the cluster and immediately rejoin, which is
useless and confusing to users. See #1820.
Signed-off-by: Sage Weil <sage@newdream.net>
The ceph_mon.cc main() will delete mon when the msgr dispatch thread
completes. Make sure we unlock before we shut down the messenger, and
avoid touching this after messenger->shutdown().
Fixes: #2090
Signed-off-by: Sage Weil <sage.weil@dreamhost.com>
