INDUSTRY

The GRIDS Center Builds Team is pleased to announce the availability of an update to the GRIDS Center Software Suite in the NMI-R5 series. This update includes: - updated versions of six components - one new component - three new platforms - the deprecation of one platform. Below is a copy of the release notes for the full details. Please note that they are considering removing support for Solaris 8 and AIX 5.1 for NMI-R6 scheduled for the October-2004 timeframe. If you have a requirement for these platforms (or any others that they currently do not support) they would greatly appreciate hearing from you. http://www.nsf-middleware.org/NMIR5/grids-center/download.asp ====================================================================== Release Notes - NMI 5.1 ======================================================================= Changes since NMI 5.0 --------------------- - Added support for Solaris 8, Solaris 9 and SuSE 8.1 on amd64 (Opteron). - Dropped support for Redhat 7.2 on ia64. - Globus has been updated to 3.2.1 to address authorizing remote party prior to doing delegation problem. [1][2] - GSI-OpenSSH has been updated to 3.4 to address server authentication logic error when a user is in the grid-mapfile, but SSH logins disabled. [3] - Condor was updated to 6.6.6 - UberFTP has been updated to version 1.9 - NWS has been updated to version 2.10.1 - MPICH-G2 has been updated to version 1.2.6 - Environment and packaging fixes for Inca - Added PBS XML Accounting Toolkit component Component notes --------------- - MPICH-G2 will not work with without two update packages from Globus. GT 3.2 moved it implementation of GlobusIO atop of the new Globus XIO. This introduced a bug in GT which causes MPICH-G2 to hang. Get the updated packages from the Globus web site (www-unix.globus.org/toolkit/advisories.html) The two packages are identified as "globus_io-5.5" and "globus_nexus-6.5" (both say "for mpich-g2" in their descriptions). - GPT-3.2 only contains some bug fixes which address gpt-build. It contains a version of the globus_core package (3.14) which only works with version 3.2 or later of the Globus Toolkit. Older versions of the toolkit should be built with GPT-3.1. - GridPort3 is not shipped for the AIX platforms. - Globus MDS2 does not work on AIX. See http://bugzilla.globus.org/globus/show_bug.cgi?id=150. The globus-information-services-server source bundle can be built on top of NMI-R5 and relinked with the script in the bug report above to get it to work. - The release is built with a mix of as many flavors as possible to provide a nice environment to build other software against. This includes the AIX bits being mostly 64bit, with the exception of the GridFTP server, which is 32bit. - Globus MMJFS does not work on AIX, and is not included in the bundles. - GSI-OpenSSH is built with "--with-tcp-wrappers" option on AIX, and with "--with-pam --with-md5-passwords --with-tcp-wrappers" options on Linux platforms. On AIX, GSI-OpenSSH is built with 32 bit. - Datacutter/STORM is built 32 bit on AIX. - The GridConfig Tools does not provide configuration for OGSA part of GT3. - Condor - 6.6.6 release notes * Opteron bundle for Suse ES 8 contains dynamically linked Redhat 7.2 binaries. * A condor_dagman job will fail and report a cycle in the Dag when XML logs are used in a single or multiple log formats. The Post Script completion event does not get converted to XML and Dagman never sees them complete or fail because of the format of the event. * If you are using Solaris 9 with LDAP as a name lookup service, Condor will not function, at least in some cases. We don't yet know if this bug affects other Solaris revisions. * Support for new proxy format. * Fixed a bug where the environment of jobs in any universe could be corrupted. * Fixed a problem where the condor_starter could crash if the job it was running used Condor's file transfer mechanism and the full path names to the job's files became longer than a few hundred characters. * The condor_startd now properly cleans up execute directories on root-squashed NFS mounts. * Fixed a memory leak in the condor_gridmanager. * Fixed a minor bug in the file transfer mechanism. Specifically, if a VANILLA job had when_to_transfer_output set to ON_EXIT_OR_EVICT, wrote more than one output file, and was actually evicted, the condor condor_shadow would have a fatal run-time error (shadow exception) and your job would be rerun. * Fixed a bug which could result in a daemon crashing while it was writing to its logfile. * The -analyze option to condor_q is only meaningful for certain universes. condor_q now warns if the output might not be meaningful. * Previously, if a scheduler universe job died with a signal, the condor_schedd would write multiple (conflicting) events into the UserLog file: a terminate event and an abort event. Now, only the terminate event is written, not the abort event. * Fixed a minor bug where if the condor_schedd crashed or was killed at just the wrong moment while a job was being removed because the periodic_remove expression had evaluated to TRUE, the job might have been successfully removed but the RemoveReason attribute could have been lost. Now, both actions are taken together atomically. If a job is successfully removed, it will always have a RemoveReason attribute. * Several other bug fixes and new features are available. Detailed list of changes from v6.6.5 to v6.6.6 can be found at http://www.cs.wisc.edu/condor/manual/v6.6.6/8_2Stable_Release.html - Datacutter/Storm Porting for Solaris platforms 8 and 9 has not been finished, this software will not be released for solaris platforms. - SRB client programs that were released with NMI R5 for 64-bit SuSe8 give segmentation fault. The problem has been fixed in R5.1. People who installed all bundle from NMI R5 and use SRB on this platform will need to reinstall SRB client issued in NMI R5.1. - NWS 2.10.1 release notes: - The NWS slapd, the part that allows to access NWS data using MDS, isn't functional on AIX and Solaris for this release. All other NWS components are fully functional; - Fixed various bugs and memory leaks: in particular fixed a bug that caused random crashed on some architectures; - Tuned the NWS internal adaptive timeout discovery; - The nws_memory now registers the series with the nws_nameserver: this avoid to have unreachable data; - Reimplemented the clique protocol: it is now more resilient and consistent even with a large number of hosts or with unreliable hosts/network; - nws_extract can now deal with hosts that are given without domain name; - Added new command, nws_ctrl, to access new (and old) NWS funcitonality. In particular this new capabilities have been implemented: - instructs a nws_sensor to register with a new nws_memory; - instructs a nws_sensor to execute a specific skill without having to start an activity; - add/remove a nws_sensor from a running clique. - Uberftp updates since NMI 5.0: * Version 1.9: Adds a fix for tcpbuf. * Version 1.8: 'tcpbuf ' will set the TCP buffer size. * Version 1.7: ALLO support was added. username/password was removed. The -H option on the commandline or to open/lopen is optional so users can now specify 'uberftp mss.ncsa.uiuc.edu' which connects to mass storage using GSI on port 2811. 15 minute timeouts form the Globus gridftp server are now handled correctly. * Version 1.6: UberFtp now defaults to passive streams mode to avoid firewall issues. If you issue 'parallel x' where x >= 2, or perform third-party transfers, the client will use an active data connection. It will also failover to active if the service denies passive mode. This change will also be more efficient then the previous versions where it used extended block mode on a single data channel. References: [1] http://bugzilla.globus.org/globus/show_bug.cgi?id=1774 [2] http://www-unix.globus.org/toolkit/advisories.html [3] http://grid.ncsa.uiuc.edu/ssh/implicitlogin.adv