On 10/31/16 05:56, xxhdx1985126 wrote:
Hi, everyone.
Recently, I deployed a ceph cluster manually. And I found
that, after I start the ceph osd through "/etc/init.d/ceph -a
start osd", the size of the log file "ceph-osd.log" is 0, and
its owner isnot "ceph" which I configured in
/etc/ceph/ceph.conf but the user who actually run the
/etc/init.d/ceph script. I read the /etc/init.d/ceph script,
and found that the command "ceph-conf" is run by the current
user with the arguments "-n $type.$id", which makes it create
a ceph-osd.log which is owned by the current user.
How should I deal with this problem? Thank you:-)
Same problem and more (/var/run/ceph and journal disk) here since
upgrading to jewel... for now I have written a small (highly
non-standards compliant) init script for some problems:
/etc/init.d/bc-ceph-init
#!/bin/bash
mkdir -p /var/run/ceph
chown ceph:ceph /var/run/ceph
chgrp -R ceph /var/log/ceph
for d in /dev/ssd*/journal_*; do
d=$(readlink -f "$d")
chown ceph:ceph "$d" #this won't survive a hot-replace
done
chmod +rx /etc/init.d/bc-ceph-init
ln -s /etc/init.d/bc-ceph-init /etc/rc2.d/S19bc-ceph-init
|
_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com