RecoverCvmfs
Procedure when CVMFS fails in a UI
- Identify the 4 CVMFS processes
[root@m3 ~]# ps ax|grep cvmfs 19206 ? S 0:00 /usr/bin/cvmfs2 <u>cachemgr</u> . 8 9 8388608000 4194304000 0 3 -1 : 19208 ? S 0:00 /usr/bin/cvmfs2 <u>cachemgr</u> . 8 9 8388608000 4194304000 0 3 -1 : 19211 ? Sl 0:00 /usr/bin/cvmfs2 -o rw,fsname=cvmfs2,allow_other,grab_mountpoint,uid=105,gid=108 cms.cern.ch /cvmfs/cms.cern.ch 19215 ? S 0:00 /usr/bin/cvmfs2 -o rw,fsname=cvmfs2,allow_other,grab_mountpoint,uid=105,gid=108 cms.cern.ch /cvmfs/cms.cern.ch
- Kill them
[root@m3 ~]# kill -9 19206 19208 19211 19215
- CVMFS should work after few seconds
[root@m3 ~]# ls /cvmfs/cms.cern.ch bin cmsset_default.csh SITECONF bootstrap.sh cmsset_default.sh slc5_amd64_gcc434 bootstrap.sh.old common slc5_amd64_gcc461 bootstrap.sh.old2 COMP slc5_amd64_gcc462 bootstrap_slc5_amd64_gcc434.log crab slc5_amd64_gcc470 bootstrap_slc5_amd64_gcc461.log crab3 slc5_amd64_gcc472 bootstrap_slc5_amd64_gcc462.log cvmfs slc5_amd64_gcc481 bootstrap_slc5_amd64_gcc470.log cvmfs-cms.cern.ch-updates slc5_ia32_gcc434 bootstrap_slc5_amd64_gcc472.log etc slc6_amd64_gcc472 bootstrap_slc5_amd64_gcc481.log rpmdb-slc5_ia32_gcc434-bkp-201102.tgz slc6_amd64_gcc480 bootstrap_slc5_ia32_gcc434.log rpmdb-slc5_ia32_gcc434-bkp-201302.tgz slc6_amd64_gcc481 bootstraptmp share var
When the above fails
If there is no cvmfs daemon running, you might find the following message in /var/log/messages
failed to initialize root file catalog (16 - file catalog failure)
In that case, you can restore the system with a reload of the cvmfs config:
cvmfs_config reload -c
After this step, you might have to restart autofs though. Template:TracNotice