On Thu, 2003-11-20 at 19:59, Daniel Veillard wrote: > On Thu, Nov 20, 2003 at 08:47:36PM +0300, Artem B. Bityuckiy wrote: > > I often heard word "smart" about up2date. The example from reality: In > > RH9 up2date downloaded new packages and offer KDE user to update system. > > User answers - yes. After update system can't work and can't boot. I > > don't know exactly what happened, but up2date tried to update glibc and > > hanged on this - and this is normal because libc is used everywhere. Why > > up2date is "smart" if it is so stupid to update glibc while KDE and a > > lot of other applications are running? > > Live update of glibc are possible and work even if the library > is in use. Qualifying the program as "stupid" because it failed > in your specific case just show you don't understand what going on. > I think millions of glibc live update have been done via RPM > be it though up2date or others update mechanism. If the RPM transaction > fails, that's bad, that probably not directly related to rpm itself > or the fact that it was a glibc update. But since there is aboslutely > no informations in your report which may help explain or reproduce the > problem, I think no further work can be done about your problem. To me the "tried to update glibc and hanged on this" sounds like a locked/hung rpm database, especially since we're talking about RHL9 whose original rpm version is a bit touchy to ctrl-c. Sure enough if one reboots the system in a middle of unfinished operation, especially when updating such a core compotent as glibc .. well, no wonder it got trashed. - Panu -
