Re: Some thoughts on the ARM build daemons
Colin Tuckley (22/01/2010): > Your list of buildd machines is seriously out of date if you are > talking about armel. You do realize you're answering to a mail dated “08 May 2007”, right? > -- > Colin Tuckley | +44(0)1223 293413 | PGP/GnuPG Key Id > Debian Developer | +44(0)7799 143369 | 0x1B3045CE > > I'm not as dumb as you look. I'm not as outdated as you look. (:p) Mraw, KiBi. signature.asc Description: Digital signature
Re: Some thoughts on the ARM build daemons
Argh! Ignore that - I was replying to an ancient message by mistake. Colin -- Colin Tuckley | +44(0)1223 293413 | PGP/GnuPG Key Id Debian Developer | +44(0)7799 143369 | 0x1B3045CE I'm not as dumb as you look. -- To UNSUBSCRIBE, email to debian-project-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Re: Some thoughts on the ARM build daemons
Aurelien Jarno wrote: > The problems mainly comes from the build daemons. Only *3 out of 7* are > building packages, and one of the three is also building stable-security > from time to time. Are you talking about the deprecated "arm" arch or the current "armel" arch? Your list of buildd machines is seriously out of date if you are talking about armel. > I think it is time to changes things. Our faster build daemons have a > 233MHz CPU with 256MB of RAM, while there are way faster ARM CPU today. I think these are arm buildds, most of the armel buildd machines are 700MHz Thecus N2100 boxes. > What about buying new ARM hardware with that money? There are new faster buildd machine being built at the moment - but they are for armel not arm. > Any comments? I suspect that people have stopped caring about arm. Colin -- Colin Tuckley | +44(0)1223 293413 | PGP/GnuPG Key Id Debian Developer | +44(0)7799 143369 | 0x1B3045CE I'm not as dumb as you look. -- To UNSUBSCRIBE, email to debian-project-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Re: Some thoughts on the ARM build daemons
Lennert Buytenhek wrote: > (I also don't think that finding ARM silicon vendors willing to donate > some man-power to the Debian ARM port would be very hard, either.) That's always good and helpful for kernel and toolchain work. Having some of these people "around" and jump in should the need arise, would be a good thing. Also, improving the arm port and installer to make it even fitter for embedded use and embedded hardware that may not be too familiar may be an appropriate goal as well. Regards, Joey -- A mathematician is a machine for converting coffee into theorems. Paul Erdös Please always Cc to me when replying to me on the lists. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Some thoughts on the ARM build daemons
Moritz Muehlenhoff wrote: > Aurelien Jarno wrote: > > - tofee: up, building packages, sometimes stable-security. > > > > I think it is time to changes things. Our faster build daemons have a > > 233MHz CPU with 256MB of RAM, while there are way faster ARM CPU today. > > How much faster is the fastest available ARM CPU compared to toffee? The 1.2GHZ dual core CPU is the Intel IOP342 (IOP341 for single core). Intel also produces a "Customer Reference Board" named IQ81342EX which is an ATX board with an IOP342. http://www.intel.com/design/iio/iop341_42.htm and about http://download.intel.com/design/iio/prodbref/31503301.pdf Regards, Joey -- A mathematician is a machine for converting coffee into theorems. Paul Erdös -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Some thoughts on the ARM build daemons
On Tue, May 08, 2007 at 07:51:49PM -0300, Henrique de Moraes Holschuh wrote: > > > > I think it is time to changes things. Our faster build daemons have a > > > > 233MHz CPU with 256MB of RAM, while there are way faster ARM CPU today. > > > > > > How much faster is the fastest available ARM CPU compared to toffee? > > > > As in, the fastest ARM CPU that exists in the world? As far as > > I know, that would be the 1.2 GHz Intel IOP342 (dual core.) > > Any reason why we cannot just buy two of these things, beef them up > with a lot of RAM, and set them up in different parts of the world? > It is not like the ARM platform is not an important one: it is being > used everywhere, and a lot of embedded work is done with it. Well, as said elsewhere in this thread, Bill Gatliff provided a 600 MHz ARM box to the Debian project for use as a buildd over half a year ago, but nothing has happened with that machine so far -- which is one of the reasons why I've said many times now that I don't think the hardware is the biggest problem that the Debian ARM port is facing. I don't think that finding ARM silicon vendors willing to donate lots of contemporary ARM hardware to Debian would be very hard. (I also don't think that finding ARM silicon vendors willing to donate some man-power to the Debian ARM port would be very hard, either.) cheers, Lennert -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Some thoughts on the ARM build daemons
On Wed, 09 May 2007, Lennert Buytenhek wrote: > On Tue, May 08, 2007 at 11:53:36PM +0200, Moritz Muehlenhoff wrote: > > > - tofee: up, building packages, sometimes stable-security. > > > > > > I think it is time to changes things. Our faster build daemons have a > > > 233MHz CPU with 256MB of RAM, while there are way faster ARM CPU today. > > > > How much faster is the fastest available ARM CPU compared to toffee? > > As in, the fastest ARM CPU that exists in the world? As far as > I know, that would be the 1.2 GHz Intel IOP342 (dual core.) Any reason why we cannot just buy two of these things, beef them up with a lot of RAM, and set them up in different parts of the world? It is not like the ARM platform is not an important one: it is being used everywhere, and a lot of embedded work is done with it. -- "One disk to rule them all, One disk to find them. One disk to bring them all and in the darkness grind them. In the Land of Redmond where the shadows lie." -- The Silicon Valley Tarot Henrique Holschuh -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Some thoughts on the ARM build daemons
Lennert Buytenhek wrote: > > > - tofee: up, building packages, sometimes stable-security. > > > > > > I think it is time to changes things. Our faster build daemons have a > > > 233MHz CPU with 256MB of RAM, while there are way faster ARM CPU today. > > > > How much faster is the fastest available ARM CPU compared to toffee? > > As in, the fastest ARM CPU that exists in the world? As far as > I know, that would be the 1.2 GHz Intel IOP342 (dual core.) Are these publicly available and affordable? If so, let's replace the existing seven buildds with two IOP342; it'll reduce the administrative overhead for DSA/buildd admins and lower the peak time for arm security builds. Right now toffee is six times slower than the second-slowest Etch archs (mipsel/powerpc) and about 32 times slower than the fastest one (s390). (All accounted for a recent xfree86 oldstable-security build). Cheers, Moritz -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Some thoughts on the ARM build daemons
On Tue, May 08, 2007 at 11:53:36PM +0200, Moritz Muehlenhoff wrote: > > - tofee: up, building packages, sometimes stable-security. > > > > I think it is time to changes things. Our faster build daemons have a > > 233MHz CPU with 256MB of RAM, while there are way faster ARM CPU today. > > How much faster is the fastest available ARM CPU compared to toffee? As in, the fastest ARM CPU that exists in the world? As far as I know, that would be the 1.2 GHz Intel IOP342 (dual core.) cheers, Lennert -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Some thoughts on the ARM build daemons
On Tue, May 08, 2007 at 11:53:36PM +0200, Moritz Muehlenhoff wrote: > Aurelien Jarno wrote: > > - tofee: up, building packages, sometimes stable-security. > > > > I think it is time to changes things. Our faster build daemons have a > > 233MHz CPU with 256MB of RAM, while there are way faster ARM CPU today. > > How much faster is the fastest available ARM CPU compared to toffee? 667MHz at least (from Intel). Maybe higher from other vendors. Hamish -- Hamish Moffatt VK3SB <[EMAIL PROTECTED]> <[EMAIL PROTECTED]> -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Some thoughts on the ARM build daemons
Aurelien Jarno wrote: > - tofee: up, building packages, sometimes stable-security. > > I think it is time to changes things. Our faster build daemons have a > 233MHz CPU with 256MB of RAM, while there are way faster ARM CPU today. How much faster is the fastest available ARM CPU compared to toffee? Cheers, Moritz -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Some thoughts on the ARM build daemons
On Tue, May 08, 2007 at 12:02:50PM +0100, Wookey wrote: > I'm not sure money is the issue - it appears to be DSA set-up time. > New hardware has been offered, many months ago, and is there, ready, > online, but (so far as I can tell) it has not been brought into use by > the people with the power to do it (DSA and arm build-admin). > > http://lists.debian.org/debian-arm/2006/11/msg8.html > > I don't know who is waiting for what, at this stage. That reminds me of sparc... where vore is dead since late 2005 (at least), but at least two live hardware offers are being ignored for months (live as in machines are online and waiting to be used). It's not like anyone really cares, as long as the remaining machines provide the modicum of functionality, but you'd expect at least the courtesy of a response. Oh, wait, no, you can't really allow yourself to expect a response, because that would break a long-time pattern of d-a member behaviour. And we can't have such silly things happen, can we! -- 2. That which causes joy or happiness. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Some thoughts on the ARM build daemons
Wookey: Wookey wrote: Bill, elmo - what happened after emails in feb? What are we waiting for now? Can anyone on this list help? I'm still here, and I think that emails aren't getting eaten anymore. b.g. -- Bill Gatliff [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Some thoughts on the ARM build daemons
Aurelien Jarno a écrit : > Wookey a écrit : >> On 2007-05-08 12:05 +0200, Aurelien Jarno wrote: >>> Hi all, >>> >>> The ARM port is getting bad [1], the percentage of packages built >>> staying a bit more than 90% for 2 weeks. Also this is confirmed by a >>> message on #debian-arm this morning: >>> >>> 09:30 < doko> please could somebody care about the python2.4, python2.5, >>> binutils, gcc-4.1 and gcj-4.1 builds for arm? for gcj-4.1, please see >>> the instructions posted on debian-ports/debian-gcc >>> >>> gcj-4.1 has to be bootstrapped manually (after python is installable >>> again), but all other packages have not been built on ARM. They are >>> blocked by python being uninstallable, but python2.4 is on position 115! >>> in the list of packages to build. >>> >>> The problems mainly comes from the build daemons. Only *3 out of 7* are >>> building packages, and one of the three is also building stable-security >>> from time to time. >>> >>> Here are the states and reasons I have been able to found on the web and >>> on IRC: >>> - grieg: up, building packages >>> - cats: up, building packages >>> - tofee: up, building packages, sometimes stable-security. >>> - smackdown: down, waiting for a 2.6 kernel >>> - elara: down, waiting for bug#421037 >>> - europa: down, waiting for bug#421037 >>> - netwinder: down, waiting for bug#421037, but seems to be unaffected >>> by this bug >>> >>> Note that the bug#421037 may be due to a too old kernel, but the glibc >>> maintainers are lacking information. >> I tried to reproduce this on my netwinder but it simply didn't happen >> here. >> >>> The situation is lasting for some time now, and I see no sign that it >>> will change soon. If the person in charge of the ARM build daemons is >>> short on time (which I can understand, life is life), why not delegate a >>> person to fix this problem? >> I would love to see Aurelien added to the arm buildd team - he has >> shown significant activity and interest in this over the last year, >> and clearly knows what is going on. >> >>> Also the build power has always been tight on this architecture, so even >>> if they are all restarted today, it would take weeks to get back to a >>> normal state. >>> >>> I think it is time to changes things. Our faster build daemons have a >>> 233MHz CPU with 256MB of RAM, while there are way faster ARM CPU today. >>> >>> Quoting our new DPL's platform: >>> >>> "So we still have that money, and I would like to use it at least to >>>fix our broken hardware." >>> >>> What about buying new ARM hardware with that money? >> I'm not sure money is the issue - it appears to be DSA set-up time. >> New hardware has been offered, many months ago, and is there, ready, >> online, but (so far as I can tell) it has not been brought into use by >> the people with the power to do it (DSA and arm build-admin). >> >> http://lists.debian.org/debian-arm/2006/11/msg8.html >> >> I don't know who is waiting for what, at this stage. There was some >> activity on this in mid-Feb when Elmo and Bill exchanged mail via me, >> after discovering that elmo's mail to bill was getting eaten somewhere >> along the way (which obviously doesn't help). They arranged to meet on >> IRC to get things sorted out. > > That's why I proposed to replace the current machines by new ones. The > local admin can just take the old disk (or copy the system if the disk > format is different) and install a kernel for the new subarch. That does > not change the IP, the account, the SSH keys, etc., so there is nothing > to ask to DSA or whoever. > >> I can also offer a reasonably fast build machine, but there didn't >> seem much point until Bills was put into use. >> >> So, yes - lets get fatrer hardware - it is now available, but until we >> can work out why we have been unable to bring hedges into use in >> nearly 6 months, and fix that problem, then money for hardware will >> not help. >> >> Bill, elmo - what happened after emails in feb? What are we waiting >> for now? Can anyone on this list help? >> > > If elmo is too busy (something that happens to everybody and that I > understand, life is life), why can't this be done by somebody else? > Don't say me elmo is not the only person able to setup a build daemon. > ^^^ just remove the not :) -- .''`. Aurelien Jarno | GPG: 1024D/F1BCDB73 : :' : Debian developer | Electrical Engineer `. `' [EMAIL PROTECTED] | [EMAIL PROTECTED] `-people.debian.org/~aurel32 | www.aurel32.net -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Some thoughts on the ARM build daemons
Wookey a écrit : > On 2007-05-08 12:05 +0200, Aurelien Jarno wrote: >> Hi all, >> >> The ARM port is getting bad [1], the percentage of packages built >> staying a bit more than 90% for 2 weeks. Also this is confirmed by a >> message on #debian-arm this morning: >> >> 09:30 < doko> please could somebody care about the python2.4, python2.5, >> binutils, gcc-4.1 and gcj-4.1 builds for arm? for gcj-4.1, please see >> the instructions posted on debian-ports/debian-gcc >> >> gcj-4.1 has to be bootstrapped manually (after python is installable >> again), but all other packages have not been built on ARM. They are >> blocked by python being uninstallable, but python2.4 is on position 115! >> in the list of packages to build. >> >> The problems mainly comes from the build daemons. Only *3 out of 7* are >> building packages, and one of the three is also building stable-security >> from time to time. >> >> Here are the states and reasons I have been able to found on the web and >> on IRC: >> - grieg: up, building packages >> - cats: up, building packages >> - tofee: up, building packages, sometimes stable-security. >> - smackdown: down, waiting for a 2.6 kernel >> - elara: down, waiting for bug#421037 >> - europa: down, waiting for bug#421037 >> - netwinder: down, waiting for bug#421037, but seems to be unaffected >> by this bug >> >> Note that the bug#421037 may be due to a too old kernel, but the glibc >> maintainers are lacking information. > > I tried to reproduce this on my netwinder but it simply didn't happen > here. > >> The situation is lasting for some time now, and I see no sign that it >> will change soon. If the person in charge of the ARM build daemons is >> short on time (which I can understand, life is life), why not delegate a >> person to fix this problem? > > I would love to see Aurelien added to the arm buildd team - he has > shown significant activity and interest in this over the last year, > and clearly knows what is going on. > >> Also the build power has always been tight on this architecture, so even >> if they are all restarted today, it would take weeks to get back to a >> normal state. >> >> I think it is time to changes things. Our faster build daemons have a >> 233MHz CPU with 256MB of RAM, while there are way faster ARM CPU today. >> >> Quoting our new DPL's platform: >> >> "So we still have that money, and I would like to use it at least to >>fix our broken hardware." >> >> What about buying new ARM hardware with that money? > > I'm not sure money is the issue - it appears to be DSA set-up time. > New hardware has been offered, many months ago, and is there, ready, > online, but (so far as I can tell) it has not been brought into use by > the people with the power to do it (DSA and arm build-admin). > > http://lists.debian.org/debian-arm/2006/11/msg8.html > > I don't know who is waiting for what, at this stage. There was some > activity on this in mid-Feb when Elmo and Bill exchanged mail via me, > after discovering that elmo's mail to bill was getting eaten somewhere > along the way (which obviously doesn't help). They arranged to meet on > IRC to get things sorted out. That's why I proposed to replace the current machines by new ones. The local admin can just take the old disk (or copy the system if the disk format is different) and install a kernel for the new subarch. That does not change the IP, the account, the SSH keys, etc., so there is nothing to ask to DSA or whoever. > I can also offer a reasonably fast build machine, but there didn't > seem much point until Bills was put into use. > > So, yes - lets get fatrer hardware - it is now available, but until we > can work out why we have been unable to bring hedges into use in > nearly 6 months, and fix that problem, then money for hardware will > not help. > > Bill, elmo - what happened after emails in feb? What are we waiting > for now? Can anyone on this list help? > If elmo is too busy (something that happens to everybody and that I understand, life is life), why can't this be done by somebody else? Don't say me elmo is not the only person able to setup a build daemon. Bye, Aurelien -- .''`. Aurelien Jarno | GPG: 1024D/F1BCDB73 : :' : Debian developer | Electrical Engineer `. `' [EMAIL PROTECTED] | [EMAIL PROTECTED] `-people.debian.org/~aurel32 | www.aurel32.net -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Some thoughts on the ARM build daemons
On Tue, May 08, 2007 at 12:05:01PM +0200, Aurelien Jarno wrote: > The ARM port is getting bad [1], the percentage of packages built > staying a bit more than 90% for 2 weeks. As to my thoughts about the Debian ARM port: I think there are more than enough people who care about the ARM port and want to help, but there doesn't seem to be a way for those people to do the work that needs doing. I.e. submitting bugs and providing patches aren't going to fix the issues that the Debian ARM port currently has. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Some thoughts on the ARM build daemons
On 2007-05-08 12:05 +0200, Aurelien Jarno wrote: > Hi all, > > The ARM port is getting bad [1], the percentage of packages built > staying a bit more than 90% for 2 weeks. Also this is confirmed by a > message on #debian-arm this morning: > > 09:30 < doko> please could somebody care about the python2.4, python2.5, > binutils, gcc-4.1 and gcj-4.1 builds for arm? for gcj-4.1, please see > the instructions posted on debian-ports/debian-gcc > > gcj-4.1 has to be bootstrapped manually (after python is installable > again), but all other packages have not been built on ARM. They are > blocked by python being uninstallable, but python2.4 is on position 115! > in the list of packages to build. > > The problems mainly comes from the build daemons. Only *3 out of 7* are > building packages, and one of the three is also building stable-security > from time to time. > > Here are the states and reasons I have been able to found on the web and > on IRC: > - grieg: up, building packages > - cats: up, building packages > - tofee: up, building packages, sometimes stable-security. > - smackdown: down, waiting for a 2.6 kernel > - elara: down, waiting for bug#421037 > - europa: down, waiting for bug#421037 > - netwinder: down, waiting for bug#421037, but seems to be unaffected > by this bug > > Note that the bug#421037 may be due to a too old kernel, but the glibc > maintainers are lacking information. I tried to reproduce this on my netwinder but it simply didn't happen here. > The situation is lasting for some time now, and I see no sign that it > will change soon. If the person in charge of the ARM build daemons is > short on time (which I can understand, life is life), why not delegate a > person to fix this problem? I would love to see Aurelien added to the arm buildd team - he has shown significant activity and interest in this over the last year, and clearly knows what is going on. > Also the build power has always been tight on this architecture, so even > if they are all restarted today, it would take weeks to get back to a > normal state. > > I think it is time to changes things. Our faster build daemons have a > 233MHz CPU with 256MB of RAM, while there are way faster ARM CPU today. > > Quoting our new DPL's platform: > > "So we still have that money, and I would like to use it at least to >fix our broken hardware." > > What about buying new ARM hardware with that money? I'm not sure money is the issue - it appears to be DSA set-up time. New hardware has been offered, many months ago, and is there, ready, online, but (so far as I can tell) it has not been brought into use by the people with the power to do it (DSA and arm build-admin). http://lists.debian.org/debian-arm/2006/11/msg8.html I don't know who is waiting for what, at this stage. There was some activity on this in mid-Feb when Elmo and Bill exchanged mail via me, after discovering that elmo's mail to bill was getting eaten somewhere along the way (which obviously doesn't help). They arranged to meet on IRC to get things sorted out. I can also offer a reasonably fast build machine, but there didn't seem much point until Bills was put into use. So, yes - lets get fatrer hardware - it is now available, but until we can work out why we have been unable to bring hedges into use in nearly 6 months, and fix that problem, then money for hardware will not help. Bill, elmo - what happened after emails in feb? What are we waiting for now? Can anyone on this list help? Wookey -- Principal hats: Balloonz - Toby Churchill - Aleph One - Debian http://wookware.org/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]