I've not heard of such a tool, but speaking purely from an end-user point of view it would be invaluable. We don't have the option to blow our storage away and start again to fix such problems. Similarly we are told with newer versions of GlusterFS to stay out of the back-end brick storage and not fiddle (which is what we used to do to fix high level logical problems). An "fsck" style tool for GlusterFS that fixes up back end structures would be incredibly useful. -Dan ---------------- Dan Mons Unbreaker of broken things Cutting Edge http://cuttingedge.com.au On 26 May 2014 20:23, Niels de Vos <ndevos@xxxxxxxxxx> wrote: > Hi, > > Pranith and Xavi were discussing a bug that caused a directory symlink > under <brick>/.glusterfs/<1st-part-gfid>/<2nd-part-gfid>/<full-gfid> to > be missing. For bugs like this, it might be useful to have a tool that > checks the integrity of the .glusterfs directory structure on a brick. > > Some simple things that it could check: > - a brick has a .glusterfs directory > - the root of the brick has a 000...001 GFID > - a .glusterfs/<1st-part-gfid>/<2nd-part-gfid>/<full-gfid> should either > be a symlink (for a directory) or a hardlink (link-count >= 2) > - the trusted.gfid xattr should match the <full-gfid> > - any file not under .glusterfs should have a trusted.gfid and that GFID > is available under the .glusterfs directory > > By default, such a tool should only report issues, and not fix them. Not > all issues might be fixable by checking one brick anyway. > > Is anyone aware of such a tool? Should we provide this? > > Thanks, > Niels > _______________________________________________ > Gluster-devel mailing list > Gluster-devel@xxxxxxxxxxx > http://supercolony.gluster.org/mailman/listinfo/gluster-devel _______________________________________________ Gluster-devel mailing list Gluster-devel@xxxxxxxxxxx http://supercolony.gluster.org/mailman/listinfo/gluster-devel
