[PATCH 2/2] add basic test case for readdir_r

Signed-off-by: Brian Chrisman --- src/client/testceph.cc | 36 +++- 1 files changed, 35 insertions(+), 1 deletions(-) diff --git a/src/client/testceph.cc b/src/client/testceph.cc index 520fa75..c24cc03 100644 --- a/src/client/testceph.cc +++ b/src/client/testce

[PATCH 1/2] fix null deref when callback invoked en route from readdir_r rather than readdirplus_r

Signed-off-by: Brian Chrisman --- src/client/Client.cc |8 ++-- 1 files changed, 6 insertions(+), 2 deletions(-) diff --git a/src/client/Client.cc b/src/client/Client.cc index 8b8fcf1..dbac2c9 100644 --- a/src/client/Client.cc +++ b/src/client/Client.cc @@ -4330,8 +4330,12 @@ static int

[PATCH 0/2] fix null deref in readdir_r callback and test

_readdir_single_dirent_cb is invoked with zeroed pointers when called beneath readdir_r rather than directly readdirplus_r. Those pointers are then dereferenced in assignment. There is still a problem in readdir_r, so I extended the basic scenario in testceph.cc. Methods readdir_r and readdirplus

Re: rados pool object listing suckage

It occurs to me that, in the non-recovery case, we are relying on the readdir() guarantee that each object will be returned at most once. If that is so, why not just create a danging symlink or something from the PG directory to represent the objects that are in recovery at the moment. Because we k

[PATCH 1/2] libceph: add debugging to understand how bad msg tag is getting sent

--- net/ceph/messenger.c | 13 +++-- 1 files changed, 7 insertions(+), 6 deletions(-) diff --git a/net/ceph/messenger.c b/net/ceph/messenger.c index e15a82c..db12abc 100644 --- a/net/ceph/messenger.c +++ b/net/ceph/messenger.c @@ -502,8 +502,8 @@ static void prepare_write_message(struct

[PATCH 2/2] libceph: fix handle_timeout() racing with con_work()/try_write()

Under heavy write load, ceph OSDs get messages from clients with bad message tags. The result is that client/OSD connection stalls for a while until the connection state gets sorted out. The client-side sequence of events that has this result is as follows: Due to the heavy write load, message p

Re: rados pool object listing suckage

So it occurs to me that one thing we could do in the Objecter layer, to make this a bit more sane, is to optionally enable caching of the missing objects. Then the OSD could send out all the missing objects with a flag saying that they're missing, the Objecter could cache this list, go through the

ceph fs

Hi! Like previous, but ceph fs instead of rbd. (i.e. iozone with 4G file). [ 783.295035] ceph: loaded (mds proto 32) [ 783.300122] libceph: client4125 fsid ff352dfd-078c-e65f-a769-d25abb384d92 [ 783.300642] libceph: mon0 77.120.112.193:6789 session established [ 941.278185] libceph: msg_new

Re: rados pool object listing suckage

On Thu, 12 May 2011, Colin McCabe wrote: > On Wed, May 11, 2011 at 2:57 PM, Sage Weil wrote: > > The ability to list objects in rados pools is something we originally > > added to round out the librados interface and to support the requirements > > of radosgw (S3 and swift both let you list object

Re: rados pool object listing suckage

On Wed, May 11, 2011 at 2:57 PM, Sage Weil wrote: > The ability to list objects in rados pools is something we originally > added to round out the librados interface and to support the requirements > of radosgw (S3 and swift both let you list objects in buckets).  The > current interface is statel

Re: rbd

On Thu, 12 May 2011, Sage Weil wrote: > equation. I'm running iozone on ext3 now and not having any problems. I take it back.. I just reproduced a similar error on ext2: random random bkwd record stride

Re: rbd

Hi Fyodor, > Hi! > > Latest (git pulled) version of 2.6 kernel. Ceph - 0.27.1 > > Still troubles with rbd. Now with ocfs2 no messages in syslog, but iozone > still return error: > > #df -h > FilesystemSize Used Avail Use% Mounted on > /dev/sda1 237G 15G 210G 7% /

Re: [PATCH 0/2] Update Client to handle self directories

Applied, thanks! On Wed, 11 May 2011, Brian Chrisman wrote: > libceph lookup of the self-referencing '.' directory fails. > Patch makes Client class handle '.' specially like it does '..'. > testceph updated to check the special cases of lstat(.) and lstat(/.). > > Brian Chrisman (2): > Add an

rbd

Hi! Latest (git pulled) version of 2.6 kernel. Ceph - 0.27.1 Still troubles with rbd. Now with ocfs2 no messages in syslog, but iozone still return error: #df -h FilesystemSize Used Avail Use% Mounted on /dev/sda1 237G 15G 210G 7% / none 2.0G 164K