On Fri, May 22, 2009 at 03:14:45PM +0200, Martin Bochnig wrote: > Why can the BE not be created, what happens? May this be related to > calling the authority "dev" (before Shawn corrected me, I then kept it > running for curiosity)?
Nope, nothing to do with that. > It is not possible that it really ran out of memory, as more than > enough swap is available, so it must be a bug, which is the reason why > I write this follow-up message. A MemoryError is exactly that -- pkg ran out of memory (unless Python is lying to us for some reason). Build 111a, when it runs into this, will tell you how much memory you're actually using at the time, and not stack trace, but won't help any with memory consumption -- you're already using a version that has all the memory consumption fixes we got in for 2009.06. > What I also noticed is that IPS forces the system into its knees (as > long as it runs, before it crashes). It used to be a lot worse, whatever comfort that might be. I don't know that anyone's done the analysis installing redistributable on a small machine, though, so it's possible the reduction just wasn't enough. I'll give this a shot on a small machine today, and see if I get the same thing. > With conary I never witnessed anything like this, not remotely. As > conary and IPS are implemented in the same languages (Python and for > its backends C), it must be possible to fix IPS. Undoubtedly. For now, though, if you want to move forward, you have a couple of options: - add more swap - free up more swap (looks like you had a hundred megs or so in /tmp) - quit the window system and upgrade from console or from ssh > As you see I removed all cross-postings from CC, because only the > engineers at caiman-discuss can help here. Actually, the folks on caiman-discuss can't help much at all, except those of us on the packaging team who are subscribed. pkg-discuss is the appropriate alias (so pkg-discuss copied, and caiman-discuss bcc'ed). Danek
