I actually found something even scarier: there are two .shard directories on the bricks. One is in the root (same level as .glusterfs), and one is in a subdirectory (which is called data/host). This seems like a huge bug to me, especially because it looks as if some shards are now in .shard/ and some shards (especially the more recent ones) are in data/host/.shard/ I suspect that might have happened because I was mounting a subdirectory like this on one of the clients: host-01:/backupv01/data/host /mnt/backup glusterfs defaults,_netdev,backupvolfile-server=host-06 0 0 Is there any explanation and remedy for this pretty tricky situation? The shard files in data/host/.shard do NOT belong to the files under data/host, but to other files in a higher-level directory. Stefan ----- Original Message ----- > To: "gluster-users" <gluster-users@xxxxxxxxxxx> > Sent: Friday, 24 January, 2020 11:36:08 > Subject: Orphaned shard files > Hi, > I have a 8TB GlusterFS volume with a lot of large changing data files, sharding > is enabled because the files are rather large (several GB to ~200GB). I noticed > it was filling up more than it should. > It seems that there a many files in the .shard directory that do not correspond > to an actual file anymore. Heal and full heal do not remove those files. What > can I do? I searched the docs, but I could not find any "shard cleanup" > procedure. > Thanks, > Stefan ________ Community Meeting Calendar: APAC Schedule - Every 2nd and 4th Tuesday at 11:30 AM IST Bridge: https://bluejeans.com/441850968 NA/EMEA Schedule - Every 1st and 3rd Tuesday at 01:00 PM EDT Bridge: https://bluejeans.com/441850968 Gluster-users mailing list Gluster-users@xxxxxxxxxxx https://lists.gluster.org/mailman/listinfo/gluster-users
