Created attachment 657996 [details] vdsm logs from server 1 Description of problem: --------------------------------------- When trying to start a volume from the UI, when the brick directories are absent, volume start fails, and glusterd crashes. The following is seen in the gluster logs - --------------------------------------- patchset: git://git.gluster.com/glusterfs.git signal received: 11 time of crash: 2012-12-04 23:04:42configuration details: argp 1 backtrace 1 dlfcn 1 fdatasync 1 libpthread 1 llistxattr 1 setfsid 1 spinlock 1 epoll.h 1 xattr.h 1 st_atim.tv_nsec 1 package-string: glusterfs 3.4.0qa2 /lib64/libc.so.6[0x327e832900] /lib64/libc.so.6[0x327e8753b5] /lib64/libc.so.6[0x327e877cd8] /usr/lib64/libglusterfs.so.0(synctask_destroy+0x22)[0x35d96496d2] /usr/lib64/libglusterfs.so.0(syncenv_processor+0x20)[0x35d9649aa0] /lib64/libpthread.so.0[0x327f0077f1] /lib64/libc.so.6(clone+0x6d)[0x327e8e5ccd] --------- Version-Release number of selected component (if applicable): 2.1-qa18.el6ev How reproducible: Frequently Steps to Reproduce: 1. Create a distribute volume with bricks present on each node, for a two-node cluster. 2. Remove the brick directories from one of the servers. 3. Try to start the volume from the UI. Actual results: The Events tab reports "Could not start Gluster volume <vol-name>" and glusterd is found to crash. Expected results: glusterd should not crash Additional info: glusterfs version on each of the storage servers - glusterfs-3.4.0qa2-1.el6rhs.x86_64
Created attachment 657997 [details] vdsm logs from server 2
Created attachment 657998 [details] gluster logs from server 1
Created attachment 657999 [details] gluster logs from server 2
Can this be simulated without using the UI? i.e. Perform all operations directly using gluster cli and see if it still crashes.
This issue can be reproduced by executing the following steps from gluster CLI - 1. Create a distribute volume with bricks present on each node, for a two-node cluster. 2. Remove the brick directories from one of the servers. 3. Execute "gluster volume start <vol-name>" from the other server (the one which still has brick directories) glusterd crashes after executing the volume start command from the other server.
Clearly sounds like a bug in glusterfs. Changing the component.
Shruti, can you please check if this is fixed in 3.4.0qa4?
Checked in 3.4.0qa4, Amar. This issue is not occurring.
marking it for ON_QA with 3.4.0qa4-1 as fixed in version (as per comment#9)
Verified in glusterfs version 3.4.0qa4.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. http://rhn.redhat.com/errata/RHBA-2013-1262.html