Issues Fixed in CFEngine for SL7.3 Upgrade

Known Issues

Need to limit yum output on overnight updates so that so many Emails are not sent.

The update_dell_firmware.sh script is not working correctly on SL7, possibly because the redhat-release file is not properly re-configured for disk updates, etc.

Fixed Issues

All issues noted below should now be fixed, or at least dealt with in some satisfactory manner

/root/tools/map_sd_for_iostat.sh is broken on the SL7.3 Lustre OSS
Sample entry on SL6
ID : 35
Name : c0d211
Device Name : /dev/sdak
by-pathName : disk/by-path/pci-0000:07:00.0-scsi-0:2:35:0

Sample entry in the device lists for SL7
Device Name : /dev/sda
by-pathName : disk/by-path/sda

The Lustre SL7 output looks just like the dCache Pool server output for similar storage on both SL6 and SL7

This may be related to the lack of files in /root/zpoolFiles, in particular the vdisk_sd_map.txt file, which contains entries such as this:
[root@umdist07 ~]# head -10 zpoolFiles/vdisk_sd_map.txt
ID                                : 0
Name                              : c0d000
Device Name                       : /dev/sdb
ID                                : 1
Name                              : c0d001
Device Name                       : /dev/sdc

This file can be created by:
  • tools/map_sd_to_by-path.sh

In fact, upon examination, it can be seen that these files should be re-created, probably hourly, to make sure that they are always up to date. Otherwise there is no way to correctly map the iostat devices, should anything happen to mis-target them.

This fix applies only to the Lustre OST pools that are zfs based. dCache pools do not use zfs.

  • autofs does not start
    • Check for existence of link from /etc/systemd/system/multi-user.target.wants/autofs.service to
    • /usr/lib/systemd/system/autofs.service
    • Should now be fixed in CFE Rev 2619

The final rsync script has the wrong arguments
  • Fixed in CFE

After first boot, cf3 does not run. Must first "mkdir /dcache" so the dcache promise set won't abort the configuration.
  • Fixed in Cobbler snippet

tools/remake_xfs_fstab.sh does not work on SL7
  • Error thrown: xfs_admin V1 inodes unsupported. Please try an older xfsprogs.
    • Fixed by doing one pre-mount of the storage
  • Fixed in CFE Rev 2618

service set_readAhead start
  • Did not quite work correctly, but, the dcache pools seem to have successfully been set to "noop"
  • Fixed now

error
Promised to edit '/etc/sysconfig/yum-autoupdate', but file does not exist

error
Bulk package schedule execution failed somewhere - unknown outcome for 'dcache-plugin-xrootd-monitor.*'

error
Bulk package schedule execution failed somewhere - unknown outcome for 'dcache.*'

Note that both of these latter 2 install fine locally, so it is a matter of where they are located in the repos.
  • All three of these errors are fixed now

tools/storage_details.sh has some funky output at the end:
Note: This output shows SysV services only and does not include native
      systemd services. SysV configuration data might be overridden by native
      systemd configuration.

      If you want to list systemd services use 'systemctl list-unit-files'.
      To see services enabled on particular target use
      'systemctl list-dependencies [target]'.
  • Fixed in cfe

The final cf3 output shows SL7 issues noted elsewhere:
[root@umfs03 ~]# cf-agent -K
   error: Promised to edit '/etc/sysconfig/yum-autoupdate', but file does not exist
   error: Finished command related to promiser 'ocsinventory-agent' -- an error occurred, returned 1
   error: Bulk package schedule execution failed somewhere - unknown outcome for 'ocsinventory-agent.*'
  • Fixed in cfe

yum-autoupdate is replaced in SL7 by yum-cron
For exclusions, they have to be added like so:
[base]
exclude = kernel* owncloud* php* httpd*
  • Fixed in cfe

perl after version 3.12 (basically the end of the line for sl6) dropped ctime.pl and a few other useful tools. This one in particular is used by /root/tools/minitime.sh, so, add this rpm for the class of machines where minitime.sh is installed.
  • Add perl-Perl4-CoreLibs rpm to list of packages

Other noted issues, most fixed. See each item.
  • autofs is not enabled and started at boot time (Now fixed)
  • dcache-server service is not enabled and started at boot time (rpms not available for SL7)
    • RPM now fixed in umatlas, it worked on sl6 and sl7. Did nothing about enabling the service
  • ntpd is neither enabled nor started (hopefully fixed in cfe)
  • ipmitool is not installed (not fixed, if we want it, we can install it)
  • the time rpm did not install (added to rpm list in cfe)
  • mailx rpm did not install (added to rpm list in cfe)
  • Two rpms are needed/required by fusioninventory-agent, and it is older for SL7 (darn)
    • perl-Data-Structure-Util-0.15-1.el7.rf.x86_64.rpm (not in rpm requires)
    • perl-Net-NBName-0.26-1.el7.rf.noarch.rpm (in rpm requires)
    • These would need to be added to the umatlas repo (done)

-- BobBall - 14 Sep 2017
Topic revision: r3 - 26 Oct 2017, BobBall
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback