In some cases (rename, rmdir, subtree map), we may need journal multiple
root inodes (/, mdsdir) in one EMetaBlob. This patch modifies EMetaBlob
format to support journaling multiple root inodes.
Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
commit 1203cd2110 (mds: allow open_remote_ino() to open xlocked dentry)
makes Server::handle_client_rename() xlocks remote inodes' primary
dentry so witness MDS can open xlocked dentry. But I added remote inodes'
projected primary dentries to the xlock list. This is wrong because
projected dentries are invisible for path traverse.
Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
Commit b03eab22e4 (mds: forbid creating file in deleted directory)
is not complete, mknod, mkdir and symlink are missed. Move the ckeck
into Server::rdlock_path_xlock_dentry() fixes the issue.
Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
commit 1174dd3188 (don't retry readdir request after issuing caps)
introduced an bug that wrongly marks 'end' in the the readdir reply.
The code that touches existing dentries re-uses an iterator, and the
iterator is used for checking if readdir is end.
Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
Failure to do so will mean that we will always ack the same leader during
an election started by another monitor. This had been working so far
because we were still acking the existing leader if he was supposed to
still be the leader; or we were acking a new potentially leader; or we
would eventually fall behind on an election and start a new election
ourselves, thus resetting the previously acked leader. While this wasn't
something that mattered much until now, the timechecks code stumbled into
this tiny issue and was failing hard at completing a round because there
wouldn't be a reset before the election started -- timechecks are bound
to election epochs.
Fixes: #3854
Signed-off-by: Joao Eduardo Luis <joao.luis@inktank.com>
Use git to get RPM_RELEASE only if this is a git repo
clone and if the git command is available on the system.
Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>
Require that the pool name be passed twice along with an force option
before we irreversibly delete an entire pool of objects.
Signed-off-by: Sage Weil <sage@inktank.com>
This reverts commit c993ac9b1f.
This is too hard to test. Requiring the pool name twice along with
--yes-i-really-really-mean-it should be sufficient.
Signed-off-by: Sage Weil <sage@inktank.com>
Currently written in C on FUSE hi-level interfaces, so error reporting
could be better. No serious work done for performance. But it's
usable as it stands.
Specify -c <conf> and a mountpoint, and images show up as files in
that mountpoint. You can create new images; they'll be created
with attributes stored in xattrs:
user.rbdfuse.imagesize: default 1GB
user.rbdfuse.imageorder: default 22
user.rbdfuse.imagefeatures: default 1 (layering)
Images may be truncated or extended by rewriting. Currently
once an image is opened, it's not closed, so it can't be deleted
or changed outside of the fuse path.
Signed-off-by: Dan Mick <dan.mick@inktank.com>
Something like v1.5 of the Amazon PHP library requires the AmazonS3
constructor to be given an array of parameters rather than using
the globals. More research needs to happen, and particularly
about the v2 API, but this might solve someone's problem with
v1.5 while we do that research.
Signed-off-by: Dan Mick <dan.mick@inktank.com>
The sidebar is now about a hundred pixels wider and the fonts
are larger throughout. This works a lot better when you get
deep into the doc structure - it used to wrap horribly.
I also fixed how literals look inside .tip and .important.
Signed-off-by: Ross Turk <ross@inktank.com>
For some reason, the lookup() retry loop (for when happened to
race with a removal and grab an invalid WeakPtr) locked
the lock again. This causes the #3836 crash since the lock
is already locked. It's rare since it requires a lookup between
invalidation of the WeakPtr and removal of the WeakPtr entry.
Fixes: #3836
Backport: bobtail
Signed-off-by: Samuel Just <sam.just@inktank.com>
Kill a set of parallel methods that are using the old addr/inst-based
msgr APIs, and instead use Connection handles. This is much safer and gets
us closer to killing the old msgr API.
Signed-off-by: Sage Weil <sage@inktank.com>
Previously we were sending these maps to dead osds via their old addrs
using a new outgoing connection and setting the flags so that the msgr
would clean up. That mechanism is possibly buggy and fragile, and we can
avoid it entirely if we just reuse the existing heartbeat Connection.
Signed-off-by: Sage Weil <sage@inktank.com>
If we receive an OSDMap on the cluster connection, requeue it for the
cluster messenger, and process it there where we normally do. This avoids
any concerns about locking and ordering rules.
Signed-off-by: Sage Weil <sage@inktank.com>
This reverts commit 44dca5c8c5058acf9bc391303dc77893793ce0be.
The allowance is not only added for btrfs as of commit
e639254a0c5f8e3528fa8f2b2b451296653556bc, which makes us happy
for both non-btrfs (lower latency) and btrfs (better small io
throughput, no big stall during commit).
Signed-off-by: Sage Weil <sage@inktank.com>
We only need to adjust up the op queue limits during commit for btrfs,
because the snapshot initiation (async create) is currently
high-latency and the op queue is quiesced during that period.
This lets us revert 44dca5c, which disabled the extra allowance because
it is generally bad for non-btrfs writeahead mode.
Signed-off-by: Sage Weil <sage@inktank.com>
Otherwise we try to read the whole object in one go, which doesn't bode
well for large objects (either non-optimal or simply broken).
Signed-off-by: Sage Weil <sage@inktank.com>
Reviewed-by: Dan Mick <dan.mick@inktank.com>
Each completed operation in the transaction proves thread
liveness, a stuck thread should still trigger the timeouts.
Fixes: #3928
Backport: bobtail
Signed-off-by: Samuel Just <sam.just@inktank.com>
Implement _process overload with TPHandle argument and use
that to ping the hb map between pgs and between map epochs
when advancing a pg. The thread will still timeout if
genuinely stuck at any point.
Fixes: 3905
Backport: bobtail
Signed-off-by: Samuel Just <sam.just@inktank.com>
