On Thu, Feb 12, 2004 at 01:58:11AM +0000, Martin Michlmayr - Debian Project Leader wrote: > * Ingo Juergensmann <ij@2004.bluespice.org> [2004-02-11 23:26]: > > Having 10 machines can mean, 10 machines can fail. Having 1 machine, means > > only one machine can fail. Whereas the first seems to introduce higher > > possibility oof failures, it also means that there is a redundancy because > > of: > > 10 machines - one fail -> 10% of CPU time fails > > 1 machine - one fail -> 100% of CPU time fails > > Sure, I know that; please see what I wrote about "redundancy" in the > case of ARM. However, adding more machines infinitely is not the > great panacea either since certain downsides are associated with > adding more machines. It created more load for ftp-master, and (more > importantly) is a bigger potential for security attacks; etc. > > However, as I wrote in my previous mail, we are working on getting > more ARM, mips and mipsel hardware, either to be used as buildd > directly or to have as fallback in case it's needed. We don't try to > pile that hardware up for fun, but to make sure we have a certain > amount of redundancy and some fallback options; plus enough machines > for various porting work. Hardware redundancy is one thing, what Debian needs is people redundancy, and Debian needs it in more than just buildd admins. In Ryan's case his packages even suffer from being out of date. Take GDM for example, at one point it was out of date for over a year and recently it was out of date for over 5 months. He is obviously very overloaded... > > Hrm, not exactly as I understood his original mail. One part of his > > complain was that Ryan was some sort of unresponsive, although there > > were several postings on debian-mips requesting the building of > > qt-x11-free, but nothing happened. *This* is some sort of > > unacceptable. > > Yeah, I was quite frustrated with this as well. Oh well, there are > just many different people in Debian and we have to try a way to work > together. Some people are very communicative while others just do > their work -- and Ryan generally does his work. I remember a recent > request on debian-mips to have a package recompiled, and while Ryan > did not respond to the mail, looking at incoming.d.o showed that he > had acted upon the request. It would have been nice to also get a > short mail saying so, but then again this would take away time from > his work. Some developers do take that time to respond and others > don't. In the case of qt-x11-free, I'm not sure exactly what the > problem was. However, simply uploading a totally untested package > as Goswin did isn't ideal either (especially since someone else had > built the package already and asked for testers). Saying "oh well" about lack of communication is not something I would expect to hear from DPL! Lack of communication is just not acceptable period. People who cannot communicate do not belong in positions of power, its as important as technical knowledge if not more so. Also there was no reason to test the qt-x11-free deb it didn't fail to build it was just stuck in Needs-Build for multiple weeks, never having been tried presumably because Ryan has marked it weak-no-auto(?) or else the buildds were dead for a prolonged time. Do the buildd admins usually test all the packages they upload? They don't even upgrade known RC buggy compilers so I seriously doubt it. > > Hmmm, there was an offer to one of my subscribed Irix Mailing lists for some > > SGI machines to give away for free on a certain day in Oberhausen (Germany). > > Some mips people in Germany are looking for new hardware (in > particular Karsten who only has some old SGI), so please forward that > posting to debian-mips. > > > > (In the meantime, to make the problem worse, casals.d.o needs a new > > > kernel and cannot be used as buildd in the meantime. This should > > > hopefully be fixed soon, though.) > > > > Erm, why can't a machine be used as a buildd *and* for DDs to port/debug > > their packages? > > I didn't say it cannot. It cannot be used as buildd at the moment > because the kernel is too old. libc on mips needs a current kernel or > something like that. > > > > In summary, the currently problematic architectures are being > > > worked on. > > > > But it seems as only Ryan, James and you are know of that. For all > > others it seems as nothing would happen. It would nice to have those > > information better communicated to other people. If someone give me > > Well, James is responsible for ARM, Ryan for mips, so obviously they > know. I did mention on -mips that one mipsel buildd was down, and the > ARM boards are a recent donation. > > > But then again, it doesn't make much sense to setup new buildds when > > you have to wait some weeks until they can get w-b access. > > You mentioned that one m68k buildd got access within a day or so, so > obviously it cannot be that bad. I mentioned before that I don't see > a problem with the m68k machines you're currently building to get > access to w-b either. As to those which haven't been added yet, > perhaps you should just get the hint and use them for something else > (such as d-i). "get the hint" another statement from our DPL that the current cabal no communication status quo should be kept. I am deeply saddened. > > I would wish that we can establish a way to interact with each other > > in a productive way for the sake of the project. Maybe a new mailing > > list for all buildd admins (and related persons such as buildd > > hosters) would be nice > > I certainly think such a mailing list would be a good idea. A mailing list that is in a visible place such as lists.debian.org for people to mail buildd admins for retries and to see status updates for archs would be very nice. :) As I understand it the various archs have buildd lists now but they are hidden. Chris
Attachment:
signature.asc
Description: Digital signature