Hi all, I found a solution : to get Ceph mounted after network at boot, we need to had "_netdev" option (managed by distribution). So far, I knew that but FUSE refuses to mount it with that option ! So we also need to patch file /sbin/mount.fuse.ceph to remove _netdev option passed to FUSE.
Patch included to my mail. Is it something that can be included in Ceph official sources ? On 04/04/2014 12:33 PM, Florent B wrote: > Hi all, > > My machines are all running Debian Wheezy. > > After a few days using kernel driver to mount my Ceph pools (with > backports 3.13 kernel), I'm now switching to FUSE because of very high > CPU usage with kernel driver (load average > 35). > > But I have a problem at boot mounting. > > I use this in fstab: > > daemonize,id=mirrors1,mon_host=monitors.my-company.net,key=AQCC9zJTIEQlCBAAUPw0OqKBXm//UxwOOqPiaA==,client_mountpoint=/fs1-mirrors1 > /fs1-mirrors1 fuse.ceph noatime 0 0 > > > > Mounting was fine with kernel driver using DNS monitor address. > > But FUSE is failing with a "Name or service not known". > > How can I do to use a DNS in this case ? > > Thank you. > > > _______________________________________________ > ceph-users mailing list > ceph-users@lists.ceph.com > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
--- /sbin/mount.fuse.ceph.original 2014-04-04 15:49:54.816762417 +0200 +++ /sbin/mount.fuse.ceph 2014-04-04 15:50:46.777924315 +0200 @@ -22,5 +22,8 @@ # strip out 'noauto' option; libfuse doesn't like it opts=`echo $4 | sed 's/,noauto//' | sed 's/noauto,//'` +# strip out '_netdev' option; libfuse doesn't like it +opts=`echo $4 | sed 's/,_netdev//' | sed 's/_netdev,//'` + # go exec ceph-fuse $cephargs $2 $3 $opts
_______________________________________________ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
