Gerry, just to know more, were the files which were getting deleted while the segfault happened, created VIA the AFR translator or, either copied directly to backend or created when you mounted glusterfs with -n ? This information can help us corner if the bug is with self-heal or not. thanks, avati 2007/7/4, Gerry Reno <greno@xxxxxxxxxxx>:
Krishna Srinivas wrote: > Gerry, > > I am trying to reproduce the prolem here. Can you give me > any clues? what were you trying to do? was it a plain "rm -rf" > or were you doing simultaneous rm -rf from the same > mount point? > > I am assuming that your setup is an afr connected to four > protocol/client. > Correct me if I am wrong. > > Thanks > Krishna > Krishna, I am using the same spec files that I sent to you previously only without '-n'. The 'rm -rf' test that was running was to delete various segments of the /usr tree that I had copied over to the cluster. The 'cp -ax' went fine, I did a lot of 'mv' commands and 'du' command variations without problem. Then I started to 'rm -rf' various branches of the tree. It was on the last branch 'lib' when the failure occurred. I was 'cd' into /mnt/glusterfs3/usr and the command was 'rm -rf lib'. The branch was 1.6G in size at the start. At the time of the failure it had only removed about 200K of the branch and that's all the information I have. Everything else you already have, the console output, the core dump, the log and the spec files. Regards, Gerry _______________________________________________ Gluster-devel mailing list Gluster-devel@xxxxxxxxxx http://lists.nongnu.org/mailman/listinfo/gluster-devel
-- Anand V. Avati
