On Sat, Nov 20, 2004 at 03:11:34AM +0100, Frank Lichtenheld wrote: > I had a hard time reproducing it. I'm not sure yet, but a good trigger seems > to use aptitude (I usually take the one from sarge by upgrading there, > didn't tested it with the one from woody yet) and do the upgrade in > interactive > mode, not by calling aptitude dist-upgrade. I wasn't able yet to reproduce it > either with apt-get dist-upgrade or aptitude dist-upgrade, only in interactive > mode. Seems that the install order produced by the dist-upgrade algorithms > usually avoids the problem. @submitters: do you remember how you did the > upgrade?
I can reproduce it at any time on my machines using "apt-get dist-upgrade", clients were installed via FAI, so I can setup a test-system and extract log-data. > - Tell people to update perl first. I don't know if this works, this would > require further testing. And why we have dist-upgrade in the first place > when people can't use it :( not really an option for people like me who are admins for a lot of machines and who depend on non-interactive updates... > - Tell people to ignore the error and just let them run apt-get -f install > after it occured which IME always worked well. If it really only happens > when using aptitude interactively this may even be acceptable, but should > nevertheless our last ressort if all other solutions turn out to be > uglier :/ been there, done that: uninstalls nearly half of the system... (2G of space will be freed) -- c u henning
