IGI gLite 3.2 x86_64 Update 32/33/34/35 (SL5) - 02/02/2012
In this IGI update we take just a subset of profiles from the gLite 3.2 release. In particular:
- CREAM CE
- Torque_[sever|client|utils]
- WN
- UI
NOTES (read before you start)
VERY IMPORTANT NOTE:
Due to conflicts between new and old versions of various packages the repository ''glite-generic'' should be disabled:
# mv /etc/yum.repos.d/glite-generic.repo /etc/yum.repos.d/glite-generic.repo.disabled
# yum clean all
One of the main consequences of having the same packages but with different versions in repositories for different services/node-types is that
the installation/configuration of two or more profiles/node-types on the same machine is no more supported!!!
MAIN NEWS:
- CREAM - This patch provides:
- an updated BLAH to address some problems when SGE is used as batch system in the CREAM CE
- Known issues (details)
- First time that run BUpdaterSGE after a long time, if there are a lot of jobs in non final state (removed or completed) in the job registry, many hours can be required to synchronize BUpdaterSGE daemon with the actual status, because a qstat of every pending job is needed.
- a fix in the glite-security-lcmaps-plugins-verify-proxy package. The fix addresses a problem affecting users holding a Terena grid certificate when they submit to CREAM through WMS.
- UI - Version 3.2.11 of the glite-UI - it includes the following changes:
- GFAL/lcg_utils has been updated to version 1.11.16-3. (patch #4642). Amongst other fixes this version avoids the generation of debug log files
- DPM and LFC clients have been upgraded to version 1.8.2. (patches #5005, #5003). The clients correspond to new releases of the services. The updated clients include some bug fixes and in the case of DPM also new features. DPM 1.8.2 can use a new filesystem selection algorithm based on weights. To allow for manipulation of the weights some of the existing clients have new options. As usual older clients continue to work but manipulation of new settings is not possible.
- The FTS clients (glite-transfer-* commands) have been updated to correspond to FTS release 2.2.5. (patch #3758). The utility script glite-transfer-se-set has been added (bug #61780).
- L&B clients have been updated to correspond to L&B release 2.1.21. (patch #4623)
- Known issues:
- As with all glite-UI releases the correct way to update from a previous (RPM) based release is to use *yum groupupdate glite-UI*
- The grid-env.sh written by yaim does not work with the standard version of zsh with SL5. See the known issue page
- More details about bug fixes & features
- WN - Version 3.2.12 of the glite-WN - it includes the following changes:
- GFAL/lcg_utils has been updated to version 1.11.16-3. (patch #4642). Amongst other fixes this version avoids the generation of debug log files.
- DPM and LFC clients have been upgraded to version 1.8.2. (patches #5005, #5003). The clients correspond to new releases of the services. The updated clients include some bug fixes and in the case of DPM also new features. DPM 1.8.2 can use a new filesystem selection algorithm based on weights. To allow for manipulation of the weights some of the existing clients have new options. As usual older clients continue to work but manipulation of new settings is not possible.
- The FTS clients (glite-transfer-* commands) have been updated to correspond to FTS release 2.2.5. (patch #3758). The utility script glite-transfer-se-set has been added (bug #61780).
- L&B clients have been updated to correspond to L&B release 2.1.21. (patch #4623).
- More details about bug fixes & features
- TORQUE_[client|server|utils]
The updated EPEL5 build of torque-2.5.7-1 as compared to previous versions enables munge[1] as an inter node authentication method.
It is highly advisable that prior to upgrading to version 2.5.7-1 of this torque package that munge is installed and enabled. A munge package[2] is of course available within EPEL5.
[1] http://code.google.com/p/munge/
[2] https://admin.fedoraproject.org/community/?package=munge#package_maintenance
To enable munge on your torque cluster:
Install the munge package on your pbs_server and submission hosts in your cluster.
On one host generate a key with /usr/sbin/create-munge-key
Copy the key, /etc/munge/munge.key to your pbs_server and submission hosts on your cluster.
Start the munge daemon on these nodes.. service munge start && chkconfig munge on
Warning: New internal format for Job arrays:
TORQUE 2.5 uses a new format for job arrays. It is not backwards compatible with job arrays from version 2.3 or 2.4. Therefore, it is imperative that the system be drained of any job arrays BEFORE upgrading. Upgrading with job arrays queued or running may cause data loss, crashes, etc, and is not supported.
This update from 2.3.10 includes numerous bugfixes and enhancements. Please read the torque Change Log. The significant changes that require action prior to upgrade are however detailed above
Prerequisites
It means that all the previous updates have to been applied before the last one:
Reference gLite Update
*
gLite 3.2 x86_64 Updates:
- gLite 3.2 Update 32 - 28/09/2011
- gLite 3.2 Update 33 - 01/11/2011
- gLite 3.2 Update 34 - 08/11/2011
- gLite 3.2 Update 35 - 15/11/2011
Reference packages
Package |
Version |
ig-metapackages |
ig26_sl5 |
''ig-yaim'' |
4.0.14-7 |
''ig-yaim-core'' |
4.0.14-1 |
Instructions
Checking/Updating YUM configuration
Repository settings
Please check the instructions you find at the following link:
IGI Repository settings for gLite 3.2
yum-autoupdate settings:
As good practice is suggested to
disable the
yum-autoupdate service. Any other settings is decided at site-manager own risk!
To disable the service it is available the following script "
disable_yum.sh" (thanks Giuseppe Platania) that can be used to REALLY disable yum-autoupdate.
Updating profiles
Find your nodetype and carefully follow the corresponding
actions.
Profiles |
Metapackages* *INSTALLATION |
Nodetypes* *CONFIGURATION |
Service restart |
Partial reconf |
Total reconf |
Actions |
Notes |
CREAM |
''ig_CREAM'' ''ig_CREAM_LSF'' ''ig_CREAM_torque'' |
''ig_CREAM'' ''ig_CREAM_LSF'' ''ig_CREAM_torque'' |
X |
- |
X |
Metapackage_update Nodetype_reconfiguration Middleware_services_restart |
- |
UI |
''ig_UI'' ''ig_UI_noafs'' |
''ig_UI'' ''ig_UI_noafs'' |
- |
- |
X |
Metapackage_update Nodetype_reconfiguration |
- |
WN |
''ig_WN'' ''ig_WN_noafs'' ''ig_WN_LSF'' ''ig_WN_LSF_noafs'' ''ig_WN_torque'' ''ig_WN_torque_noafs'' |
''ig_WN'' ''ig_WN_noafs'' ''ig_WN_LSF'' ''ig_WN_LSF_noafs'' ''ig_WN_torque'' ''ig_WN_torque_noafs'' |
- |
- |
X |
Metapackage_update Nodetype_reconfiguration |
- |
Actions
Each of the following sections reports the actions that may be follow in order to upgrade your profiles. Please execute
ONLY the actions listed on the table above in the column "Actions" and in the row corresponding to your profile!
Metapackage update
Please see
yum tips for additional information about
YUM.
- Update YUM metadata and install the new rpms.
yum clean all
yum install
Please remember that on case of WN you should use the "groupinstall":
yum groupinstall ig_WN_torque
On CE CREAM:
Before installing/updating a CREAM CE remember to install xml-commons-apis:
yum update xml-commons-apis
When installing/updating a CREAM CE node, a dependency problem such as:
Error: Missing Dependency: libcares.so.0()(64bit) is needed by package glite-security-gss-2.0.0-6.sl5.x86_64
could be seen. In this case, the update should be done doing:
yum update --exclude=c-ares
Please note that doing:
yum update ig-CREAM_torque
or
yum update ig-CREAM_LSF
you are not guaranteed to have all relevant RPMs updated.
When installing/updating
ig_CREAM_torque (in general torque) remember to change the torque directory
On the Batch Server (usually CREAM CE host)
/usr/sbin/create-munge-key
scp /etc/munge/munge.key <all cluster node>
service munge start
chkconfig munge on
Editing local configuration files
* In the last updates many variables have been changed or moved to ''services'' and ''defaults'' directories. Please carefully check your ''<site-info.def>'' and files under ''services/'' and ''nodes/'' directories comparing to the new ones distributed with latest released ''yaim'' modules. For example you can use ''vimdiff'' like here below:
vimdiff /opt/glite/yaim/examples/siteinfo/ig-site-info.def
You may find more detailed information on
configuration files structure here:
and on
configuration variables here:
Nodetype reconfiguration
IMPORTANT: /services directory
Please ensure that you have all the
NEEDED configuration files under your ''
/services'' directory (for example: ''ig-se_dpm_mysql'', ''ig-se_dpm_disk'', ...).
For any details please refer to the reference "IGI YAIM configuration files”
IMPORTANT: /services directory
Please ensure that you have all the NEEDED configuration files under your ''/services'' directory (for example: ''ig-mpi'', ''ig-mpi-ce'', ''ig-mpi-wn'', ...).
Remember to add the variable TORQUE_VAR_DIR in services and change the BATCH_LOG_DIR in ig-site-info.def to /var/torque
/opt/glite/yaim/bin/ig_yaim -c -s <site-info.def> -n <nodetype>
Where '''' is one of the nodetypes listed in the "Nodetypes" column of the table above.
NOTE: Please remember to add ”-n ig_BDII_site” if you want to reconfiguring a site BDII on top of your CE
Middleware services restart
Look at your ''chkconfig'' services list and restart all middleware-related service:
chkconfig --list | grep 3:on | sort
Then restart with:
service <middleware-service> restart
-- SergioTraldi - 2012-02-01