[Gluster-users] fts_read failed

[Stefan Bergstrand]

Hi,

I have simple installation, using mostly defaults, of two mirrored
servers. One brick, one volume.
GlusterFS version is 3.12.1 (server and client). All hosts involved are
Debian 9.1.

On another host I have mounted two different directories from the
cluster using /etc/fstab:

gfs1,gfs2:/vol1/sites-available/ws0 /etc/nginx/sites-available glusterfs
defaults,_netdev 0 0

and

gfs1,gfs2:/vol1/webroots/ws0 /var/www glusterfs defaults,_netdev 0 0


I cd to /var/www/example.com/public_html/www.example.com/ and run:

chown -R auser:agroup .

and get:

chown: fts_read failed: No such file or directory


If I try the same on one of the glusterfs servers, it works fine as
expected.


However, if I run:

chown auser:agroup *
chown auser:agroup */*
chown auser:agroup */*/*
[...]

all the way to the bottom of the directory tree, it all works fine.


Also, and this is really weird: I just discovered that if I shut down
nginx, the "chown -R" works...
According to "lsof| grep /var/www" no files under that mount point are
opened.
I unmounted /etc/nginx/sites-available, and it made no difference.

How could nginx possible make chown fail like this? And only when using
"-R".

(Of course, it works on a similar host/setup apart from all files
residing on local filesystem...)

Best regards,

/Stefan Bergstrand