Daniel Burrows schrieb: > On Fri, Dec 22, 2006 at 11:13:31AM +0100, Bastian Venthur <[EMAIL PROTECTED]> > was heard to say: >> very often when I update systems wich have not been updated for a while >> (> 1 week) I notice that aptitude hangs while dowhloading the new >> Package-lists (not the packages). I have to kill aptitude and use >> apt-get update (which works fine). > > I've heard of some problems with apt-get as well in this department, but > I haven't seen it myself with either aptitude or apt-get. > > Two quick questions: > > (1) When aptitude crashes, does it repeatedly crash? (i.e., does it > crash every time you try to update, or just the first time?)
When it chrashes, it often happens that it crashes again after I killed it and start it again and try to update the package list. Usually after a few tries it just works again. Sorry I know this sounds strange but behavior like this is not unusual when concurrency is involved. > > (2) If you use apt-get instead of aptitude *before* trying aptitude > update, does it ever crash? Ok speaking of *this* bug not, since apt-get does the dirty job for aptitude. This is just a wild guess but maybe it has something to do with the time aptitude needs to get the package list. When it takes longer (eg you haven't updated it for a few weeks) the possibility increases to make aptitude hang but on a dayly updated system the bug is unlikely to appear since aptitude just needs a few seconds to finish it's job. But of course there is still #348481. > Also, it would be greatly helpful if you could compile the program with > debugging support, then -- when it hangs -- attach with gdb and get a > backtrace (preferably a backtrace of all threads -- i.e., run "info threads" > and "thread 0", "thread 1", etc, getting a backtrace for each). Ok I'll see what I can do but, since this bug is more likely to happen when you don't have updated your package list for a few days/weeks, it would be very helpful if I'm able to fake such data so I don't have to wait a week for the next test. > > Otherwise, I guess I'll just wait to see if it ever happens to me. The > updating code has been more or less the same for over a year, though, and > I have yet to see a lockup. Please note that I can't remember having seen this bug on one of my daily updated sid-boxes. But I happens quite often on my etch boxes which I update just once in a week. Cheers, Bastian -- Bastian Venthur http://venthur.de Debian Developer venthur at debian org -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
