Bugzilla – Bug 681091
OpenSuSE 11.4: unable to install from nfs-shared .iso
Last modified: 2011-08-26 12:42:11 UTC
User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.2.13) Gecko/20101230 Mandriva Linux/1.9.2.13-0.2mdv2010.2 (2010.2) Firefox/3.6.13 Hi, I used to install Opensuse on many computers using a mounted .iso file nfs-shared over my home network. Booting the "to be installed computer" with the -NET cd, pressing F4 to specify DHCP+NFS... It worked great until 11.4; when Yast starts, I'm stuck with this error message: "Unable to create repository from URL 'nfs://192.168.2.10/vm/iso_mount'. Details: Failed to mount 192.168.2.10:/vm/iso_mount on /var/adm/mount/AP_0x00000001: Unable to verify that the media was mounted" I can confirm that before starting yast, the mount is ok and that the "Installation system" is loaded over nfs from the server machine. Does someone have a hint for me ? Thanks, Richard Reproducible: Always Steps to Reproduce: 1.Server side: mount -t iso9660 -o loop,ro /locatio-of-[openSUSE-11.4-DVD-i586.iso] /[mount-dir] 2.boot host with openSUSE-11.4-NET-i586.iso cd 3.press F4 to switch to NFS source, DHCP connection, start installation Actual Results: Installation starts, displays message "Loading Installation System 1/6" ... This is loading from nfs. (I can verify in console mode that nfs is connected and I'm able to browse te shared .iso file). When Yast GUI starts (select language, click next, wait for system check) then it displays this message: "Unable to create repository from URL 'nfs://192.168.2.10/[mount-dir]'. Details: Failed to mount 192.168.2.10:/[mount-dir] on /var/adm/mount/AP_0x00000001: Unable to verify that the media was mounted" Expected Results: Proceed to Partitonning steps.
Created attachment 420546 [details] Network installation log file
It seems my message disapeared when I attached the log file, so I'll retype it. I've been trying to do an NFS PXE install and I have the same problem. Here's a few tests I did to try and figure out what didn't work. In my case, the NFS server is a solaris 10 machine. Just to be sure I tried the install from another solaris server and got the same result. Then I set up a linux NFS server and that worked. I tried this both with an ISO install and a repo install with the same results. I setup an http server on the solaris machine and that worked. I setup an http server on a linux machine that was remote NFS mounting the same directory from the solaris machine and that worked. I've been installing opensuse with this setup since opensuse 11.1 and it's the first version that causes this problem. From what I understand, there seems to be 2 steps to the installation: 1 - start the preinstall program. This proceeds to mount the NFS directory where it get the installation program. This part seems to be working correctly since I get the 1/6 ... 6/6 messages and then the GUI install starts 2 - installation program. This starts, shows the first page and when I click next it tries to add the repository. This is where it fails with the error message. So apparently the 1st mount works, but not the second. When the install fails, I can go to one of the text consoles and try to mount the directory manually and that works correctly. There's no error from the command and I can list all the files. One other thing I noticed from the ISO install is that there seems to be a « / » missing between the directory and the ISO filename. For example: nfs://server/export/suse.iso becomes server /exportsuse.iso However, making a correction doesn't fix the problem. Typical y2log file attached in the previous message.
I reassign this from screening team to maintainer. Maybe it's in the packager and so I put Ladislav into CC
I think the problem is not in the installer but in zypper. On opensuse 11.4 machine client# zypper ar -c nfs://nfs.example.com/export/suse/opensuse/distribution/11.4/repo/oss test2 Adding repository 'test2' [\] Failed to mount nfs.example.com:/export/suse/opensuse/distribution/11.4/repo/oss on /var/adm/mount/AP_0x00000001: Unable to verify that the media was mounted Abort, retry, ignore? [a/r/i/?] (a): a Failed to mount nfs.example.com:/export/suse/opensuse/distribution/11.4/repo/oss on /var/adm/mount/AP_0x00000001: Unable to verify that the media was mounted Abort, retry, ignore? [a/r/i/?] (a): ^C On opensuse 11.3 machine client# zypper ar -c nfs://nfs.example.com/export/suse/opensuse/distribution/11.4/repo/oss test1 Adding repository 'test1' [done] Repository 'test1' successfully added Enabled: Yes Autorefresh: No URI: nfs://nfs.example.com/export/suse/opensuse/distribution/11.4/repo/oss
(In reply to comment #5) > I think the problem is not in the installer but in zypper. In zypp, actually.
This happened for me on a remote system with a few repositories from a external HDD which are mounted successful on my local system and are served with nfs. Most times i have no physical access to the remote system, so i have to upgrade with nfs. I don't want to copy the repos to this system, because there only a small HDD installed. The configuration is IMHO the same as on openSUSE 11.3, used on both system before, updated with 'zypper dup', on openSUSE 11.3 this config worked without problems. Upgrades on the local system with 11.4 are ok FYI: Two folders on the remote system should be mounted with fstab, but most time i have to mount them manually as root. This also worked on 11.3. Is this possible nfs-specific? thx Bernd
I see that problem with the current factory build and think it is the same as described: (extract from zypp.log) 2011-06-12 23:50:25 <1> su113(4668) [zypp++] ExternalProgram.cc(start_program):229 Executing '/bin/mount' '-t' 'nfs' '-o' 'ro,nolock,soft,timeo=300' 'mvhl:/data2/f actory/repo/debug' '/var/adm/mount/AP_0x00000001' 2011-06-12 23:50:25 <1> su113(4668) [zypp++] ExternalProgram.cc(start_program):381 pid 4674 launched 2011-06-12 23:50:26 <1> su113(4668) [zypp++] ExternalProgram.cc(checkStatus):482 Pid 4674 successfully completed 2011-06-12 23:50:26 <1> su113(4668) [zypp++] Mount.cc(Status):260 exit code: 0 2011-06-12 23:50:26 <1> su113(4668) [zypp] Mount.cc(mount):156 mounted mvhl:/data2/factory/repo/debug /var/adm/mount/AP_0x00000001 2011-06-12 23:50:26 <1> su113(4668) [zypp++] MediaHandler.cc(checkAttached):520 Forced check of the mount table 2011-06-12 23:50:26 <1> su113(4668) [zypp] MediaHandler.cc(checkAttached):588 Looking for media(nfs<mvhl:/data2/factory/repo/debug>)attached(*/var/adm/mount/AP_0x0 0000001) 2011-06-12 23:50:26 <1> su113(4668) [zypp++] MediaHandler.cc(checkAttached):595 MountEntries: { 2011-06-12 23:50:26 <1> su113(4668) [zypp++] MediaHandler.cc(checkAttached):595 rootfs on / type rootfs (rw) 2011-06-12 23:50:26 <1> su113(4668) [zypp++] MediaHandler.cc(checkAttached):595 devtmpfs on /dev type devtmpfs (rw,relatime,size=248408k,nr_inodes=62102,mode=755 ) 2011-06-12 23:50:26 <1> su113(4668) [zypp++] MediaHandler.cc(checkAttached):595 tmpfs on /dev/shm type tmpfs (rw,relatime) 2011-06-12 23:50:26 <1> su113(4668) [zypp++] MediaHandler.cc(checkAttached):595 tmpfs on /run type tmpfs (rw,nosuid,nodev,noexec,relatime,mode=755) 2011-06-12 23:50:26 <1> su113(4668) [zypp++] MediaHandler.cc(checkAttached):595 devpts on /dev/pts type devpts (rw,relatime,gid=5,mode=620,ptmxmode=000) 2011-06-12 23:50:26 <1> su113(4668) [zypp++] MediaHandler.cc(checkAttached):595 /dev/sda7 on / type ext4 (rw,relatime,user_xattr,acl,barrier=1,data=ordered) 2011-06-12 23:50:26 <1> su113(4668) [zypp++] MediaHandler.cc(checkAttached):595 proc on /proc type proc (rw,relatime) 2011-06-12 23:50:26 <1> su113(4668) [zypp++] MediaHandler.cc(checkAttached):595 sysfs on /sys type sysfs (rw,relatime) 2011-06-12 23:50:26 <1> su113(4668) [zypp++] MediaHandler.cc(checkAttached):595 debugfs on /sys/kernel/debug type debugfs (rw,relatime) 2011-06-12 23:50:26 <1> su113(4668) [zypp++] MediaHandler.cc(checkAttached):595 securityfs on /sys/kernel/security type securityfs (rw,relatime) 2011-06-12 23:50:26 <1> su113(4668) [zypp++] MediaHandler.cc(checkAttached):595 tmpfs on /var/lock type tmpfs (rw,nosuid,nodev,noexec,relatime,mode=755) 2011-06-12 23:50:26 <1> su113(4668) [zypp++] MediaHandler.cc(checkAttached):595 tmpfs on /var/run type tmpfs (rw,nosuid,nodev,noexec,relatime,mode=755) 2011-06-12 23:50:26 <1> su113(4668) [zypp++] MediaHandler.cc(checkAttached):595 none on /proc/fs/vmblock/mountPoint type vmblock (rw,relatime) 2011-06-12 23:50:26 <1> su113(4668) [zypp++] MediaHandler.cc(checkAttached):595 mvhl:/data2/factory/repo/debug/ on /var/adm/mount/AP_0x00000001 type nfs (ro,rela time,vers=3,rsize=32768,wsize=32768,namlen=255,soft,nolock,proto=tcp,timeo=300,retrans=2,sec=sys,mountaddr=192.109.84.122,mountvers=3,mountport=895,mountproto=udp, local_lock=all,addr=192.109.84.122) 2011-06-12 23:50:26 <1> su113(4668) [zypp++] MediaHandler.cc(checkAttached):595 } 2011-06-12 23:50:26 <2> su113(4668) [zypp] MediaHandler.cc(checkAttached):606 Attached media not in mount table ...
I spended some time to explore the issue. A different behavior is found in libzypp-9.8.1/zypp/media/Mount.cc: Mount::getEntries ent.mnt_fsname returned by getmntent_r (line 305) always ends with a '/' character if mounted by nfs. Thats the reason why ref.mediaSource->equals( media) ( libzypp-9.8.1/zypp/media/MediaHandler.cc: line 559 ) will always fail. I dont know, if this a kernel thing ( nfs mount / mtab ) or can be solved by cutting off the trailing '/' in ent.mnt_fsname in Mount.cc.
Stripping the trailing '/' should do. *** This bug has been marked as a duplicate of bug 710269 ***