Hide Forgot
If glusterfs/glusterfsd is given a relative path for the vol file, it crashes in parse_opts on solaris. This is due to getcwd() being parsed a size 0. On solaris, the size has to be non-zero, whereas in linux, if it is zero, it allocates buffer of required size. The fix is to pass the size as PATH_MAX, which will be handled fine on both solaris and linux
PATCH: http://patches.gluster.com/patch/6209 in master (Fix getcwd size arg to make it work in solaris.)
bash-3.00# ls bricks info new.192.168.1.73.data-dir1.vol run cksum new-fuse.vol new.192.168.1.73.data-dir2.vol bash-3.00# glusterfsd -f new.192.168.1.73.data-dir1.vol bash-3.00# less /usr/local/var/log/glusterfs/ .cmd_log_history nfs.log bricks/ usr-local-etc-glusterfs-glusterd.vol.log etc-glusterd-vols-new-new.192.168.1.73.data-dir1.vol.log bash-3.00# less /usr/local/var/log/glusterfs/etc-glusterd-vols-new-new.192.168.1.73.data-dir1.vol.log [2011-03-11 06:54:52.411192] E [socket.c:338:__socket_server_bind] tcp.new-server: binding to failed: Address already in use [2011-03-11 06:54:52.411268] E [socket.c:341:__socket_server_bind] tcp.new-server: Port is already in use [2011-03-11 06:54:52.411306] E [xlator.c:784:xlator_init] new-server: Initialization of volume 'new-server' failed, review your volfile again [2011-03-11 06:54:52.411318] E [graph.c:331:glusterfs_graph_init] new-server: initializing translator failed [2011-03-11 06:54:52.411411] I [glusterfsd.c:710:cleanup_and_exit] glusterfsd: shutting down