Friday, 29th of March 2024 03:21 (CET)

 Evaluation of DistributionKit 8.0.4 at MPI 


The following linux versions have been installed on an old batch computer (former batch01) with dual 400 MHz CPU, 512 MB memory:

SuSE-9.0: needed gcc-3.2 in addition to the base install, libshift.so and 2 symlinks to system libs with different names
Result: O.K.

SuSE-9.1: needed gcc-3.2 in addition to the base install, libshift.so and 2 symlinks to system libs with different names
Result: FAILED

Scientific Linux Cern 3 rc2: needed libshift.so
Result: FAILED

Debian testing 2004-07-13 (by Stefan Kluth): needed gcc-3.2, g77-3.2, g++-3.2, a link from make to gmake, libmotif3, libshift.so and 2 symlinks to system libs with different names
Result: O.K.

Debian stable (by Stefan Kluth): needed gcc-3.2, g77-3.2, g++-3.2, a link from make to gmake, libmotif3, libshift.so and 2 symlinks to system libs with different names
Result: FAILED

Ubuntu 4.1 preview (by Stefan Kluth): needed gcc-3.2, g77-3.2, g++-3.2, a link from make to gmake, libmotif3, libshift.so and 2 symlinks to system libs with different names
Result: O.K.

Conclusions:

The SuSE-9.1 test was the only one with a 2.6.x linux kernel. The failure here was only due to 5 warning messages in one digitization job. Nevertheless was the output of this and the other jobs o.k.
The failure of Scientific Linux Cern 3 rc2 was much more severe as many jobs crashed with memory allocation problems in a call to mzpaw_ in libHbookCnv.so. This is most surprising as this distribution came native with the correct compiler gcc-3.2.3.
To judge Debian testing 2004-07-13 is difficult as the testing releases of Debian change daily. However it seems unlikely that the parts vital for the test change dramatically in the next few weeks ...
In fact the preview of the Debian based Ubuntu 4.1 looks promissing. The failure of Debian stable is identical to SuSE-9.1 - the same 5 warning messages cause the failure message and as in SuSE-9.1 all produced results are o.k. The SuSE-9.0: distribution is the only stable distribution in the test not failing (although SuSE-9.1 and Debian-stable came also close). My personal preference from this test would therefore be SuSE-9.0, but Debian and SuSE-9.1 would also give workable solutions for the MPI to participate in DC2.

Last modified: Friday, 1st of October 2004 13:54 (CEST) Send Comments to Menke@MPPMU.MPG.DE