[Gluster-devel] 3.4.0beta2 crash in conservative merge?

Emmanuel Dreyfus manu at netbsd.org
Sat May 25 17:32:33 UTC 2013


Vijay Bellur <vbellur at redhat.com> wrote:

> Seems to be happening when rebalance was running and there was active 
> I/O from this client. Can you check which dereference leads to this SEGV?

See my other post: it crashes on line 818, because local->fd is NULL.. Since it
was used at line 804 without crashing, it must be a race conditon.

804             ret = fd_ctx_del (local->fd, this, NULL);
805             if (!ret) {
806                     fd_unref (local->fd);
807                     ret = 0;
808                     goto out;
809             }
810     
811             /* perform open as root:root. There is window between linkfile
812              * creation(root:root) and setattr with the correct uid/gid(gdb)
813              */
814             SYNCTASK_SETID(0, 0);
815             /* if 'local->fd' (ie, fd based operation), send a 'open()' on
816                destination if not already done */
817             if (local->loc.inode) {
818                     ret = syncop_open (dst_node, &local->loc,
819                                        local->fd->flags, local->fd);

-- 
Emmanuel Dreyfus
http://hcpnet.free.fr/pubz
manu at netbsd.org




More information about the Gluster-devel mailing list