gLite 3.1
glite-WN - Update to version 3.1.22-0
|
Date |
14.11.2008 |
Priority |
Normal |
Description
VOMS
An updated documentation can be found here:
VOMS Core Services User Guide: https://edms.cern.ch/file/973684/1/voms-guide.pdf
VOMS Admin User Guide: https://edms.cern.ch/file/974094/1/voms-admin-user-guide.pdf
VOMS Installation and Configuration Guide: https://edms.cern.ch/file/974982/1/voms-installation-configuration-guide.pdf
Contains a fix for bug #38506 (Previous Predictable Order with --voms flags was lost.). Now the order of FQANs in the VOMS proxy follows the order in which FQANs are specified on the command line as values of the "-voms" options. This does _not_ apply to the shorthand "<voname>:/Role=<role>" syntax.
Developer changes:
New API added for the C and C++ APIs. VOMS_SetVerificationTim(time_t time).
As usually, except where noted, everything is backwards-compatible with previous version, both on the protocol level (i.e. client and server versions may be freely mixed) and ABI level (API libraries can be upgraded with newer versions without needing recompilation of the linking parties)
glite-SE_dcache
In order to upgrade, please rerun the yaim configuration, e.g.:
./yaim -c -s site-info.def -n glite-SE_dcache_pool
./yaim -c -s site-info.def -n glite-SE_dcache_admin_postgres
Fixed bugs:
- Error message of SrmSpaceManager command 'reserve' is misleading/useless.
- PinManager fills up log files with SQL queries pool <-> xrootd-door interaction.
- /opt/d-cache/bin/dcache script doesn't start replica manager ( upper case lower case issue)
- FTP Bug, cd (ftp CWD) does not work anymore (RT #3007) busy loop in SpaceSweeper
- gPlazma returns the same FQAN for the same DN even if the FQAN changes for subsequent requests.
- ftp clinet/server checksum format miss match
- undefied NAMESPACE did not fall back to pnfs
- 'rc onerror fail' doesn't work properly.
- gPlazma web interface to verify correct configuration.
- gsidcap failes if connected to two different doors.
- high pnfs load on srmrm
- Cannot open files with PROOF
- TChain from ROOT not working properly against xrootd
- lcg-ct returns Xrootd:// instead of root://
- SrmSpaceManager : 'ls -l' doesn't report access latency and retention policy of space tokens.
- etc/node_config interpreter problem fixed
- Number of database connections seems not be limited (PinManager)
- Various fixes/improvements in installation system (rt #3082, #3160, #3252)
- File couldn't be unpinned if the disk copy didn't exist any more.
- Pin Manager is leaking db connections
- Permission handler does not respects secondary GID
- Multiple GID support in gPalzma
- globus-url-copy doesn't use the WriteToken tag to select a Space Token
- Size of removed file(s) is not returned to the token using the chimera fs backend.
- (limited) Pin Time after a bring-online start at the time when file touches the disk after a restore.
GFAL
glite-data-gfal
- Modify ETICS configuration file template to have dcache-client dependency for SLC4 and dcache-dcap for SLC5
- Experimenting with Python 2.5 support.
- Using --with-swig-prefix to really use ETICS relocated Swig dependency.
- Setting SWIG_LIB, because the one built into the 'swig' binary will not be valid in a userspace build, when Swig is relocated.
- Improve error message in case of soap error different from SOAP_EOF, and with no error string
- added missing libraries to python modules
- The proper fix was using the WITH_NOGLOBAL option instead.
- Added --with-python-location configuration option to be able to build with alternative Python versions.
- SOAP_FMAC3 and SOAP_FMAC5 are used in the generated gSOAP files, however stdsoap2.h uses SOAP_FMAC1, which has to be redefiened as well, when included in these sources.
This should fix the SL5 build.
glite-data-dm-util
- Unregister LFN in lcg_cr function when not able to register SURL in this LFN
- Print error message in lcg_rep function when not able to register SURL in the LFN
- Experimenting with Python 2.5 support.
- Solving library dependies of python module
- using --swig-prefix instead of --swig-location
- Adding 'lib64' variations to the possible SWIG_LIB locations.
- Fixed 'if' tests in configure.ac: [] are removed by m4.
- Setting SWIG_LIB, because the one built into the 'swig' binary will not be valid in a userspace build.
- using relative path to stop libtool being too smart?
- added missing python libraries
- Added --with-python-location configuration option to be able to build with alternative Python versions.
lcg-gt does not return all the information to work with dcache and this causes lcg-sd to fail. For information on how to use the lcg_utils with dCache please have a look at the following web page:
http://trac.dcache.org/trac.cgi/wiki/manuals/lcg_utils_and_dcache
glite-SE_dcache
In order to upgrade, please rerun the yaim configuration, e.g.:
./yaim -c -s site-info.def -n glite-SE_dcache_pool
./yaim -c -s site-info.def -n glite-SE_dcache_admin_postgres
Fixed bugs:
- Error message of SrmSpaceManager command 'reserve' is misleading/useless.
- PinManager fills up log files with SQL queries pool <-> xrootd-door interaction.
- /opt/d-cache/bin/dcache script doesn't start replica manager ( upper case lower case issue)
- FTP Bug, cd (ftp CWD) does not work anymore (RT #3007) busy loop in SpaceSweeper
- gPlazma returns the same FQAN for the same DN even if the FQAN changes for subsequent requests.
- ftp clinet/server checksum format miss match
- undefied NAMESPACE did not fall back to pnfs
- 'rc onerror fail' doesn't work properly.
- gPlazma web interface to verify correct configuration.
- gsidcap failes if connected to two different doors.
- high pnfs load on srmrm
- Cannot open files with PROOF
- TChain from ROOT not working properly against xrootd
- lcg-ct returns Xrootd:// instead of root://
- SrmSpaceManager : 'ls -l' doesn't report access latency and retention policy of space tokens.
- etc/node_config interpreter problem fixed
- Number of database connections seems not be limited (PinManager)
- Various fixes/improvements in installation system (rt #3082, #3160, #3252)
- File couldn't be unpinned if the disk copy didn't exist any more.
- Pin Manager is leaking db connections
- Permission handler does not respects secondary GID
- Multiple GID support in gPalzma
- globus-url-copy doesn't use the WriteToken tag to select a Space Token
- Size of removed file(s) is not returned to the token using the chimera fs backend.
- (limited) Pin Time after a bring-online start at the time when file touches the disk after a restore.
Please also have a look at the list of known issues.
This update fixes various bugs. For the full list of bugs, please see list below.
Fixed bugs
Number | Description |
#21916 |
Bad error message/behaviour with incorrect file: URL format in lcg-cr |
#33902 |
VOMS MySQL replication: adapt for 2.x schema, add encryption |
#35569 |
glite-security-voms-api-noglobus-1.8.3-3 and glite-security-voms-api-cpp-1.8.3-3 contain the same files |
#36052 |
voms-proxy-info shouldn't require AC check in all cases |
#36092 |
[VOMS 1.8.3] VOMS server log file created with bizarre permissions |
#36573 |
voms-proxy-info has bad return codes |
#37008 |
VOMS server is always returning a short FQAN |
#37071 |
VOMS api should specify in error cases whether the voms AC is expired or not yet valid. |
#37704 |
There is a possible vulnerability concerning proxies |
#38506 |
Previous Predictable Order with --voms flags was lost. |
#38824 |
VOMS doesn't build with VDT globus 1.10.1-1 because it uses system OpenSSL |
#39622 |
voms_replica_master_setup.sh forces use of SSL connections |
#39625 |
VOMS-CORE: Cannot specify two or more --voms options for the same VO in the command line |
#41094 |
VOMS-CORE: requesting an unassigned role does not return an error |
#41095 |
VOMS-CORE: request fails when requesting a role that is not assigned |
#41116 |
VOMS-CORE: -noregen functionality is broken |
#41773 |
[VOMS-CORE] No error reported when non-existent group requested |
#41831 |
segmentation fault in lcg-ls |
#42628 |
DM Python libs are not in PYTHONPATH (UI 3.1.18-0 - 3.1.20-0) + undefined symbol: globus_i_gass_copy_module |
Updated rpms
The RPMs can be updated using yum via
Service reconfiguration after update
Not needed.
Service restart after update
Not needed.
How to apply the fix
- Update the RPMs (see above)
- Update configuration (see above)
- Restart the service if necessary (see above)
|
|