On Mon, Feb 11, 2008 at 12:08:13AM +0900, Charles Plessy wrote:
> Le Sat, Feb 09, 2008 at 08:10:20PM -0500, Philippe Cloutier a écrit :
> >
> > it would be much more efficient to work on buildd
> > redundancy (or other improvements to the buildd network).
>
> By the way, is there a plan to solve the problem of mips not keeping up
> apart waiting for a miracle to happen?
>
> Sorry to be rude, but I am just so surprised that there is a such big
> problem and that apparently nothing is done. If people are working on
> the issue, just let us know, they will get many kudos and everybody will
> be happy. In the absence of any communication, you know, there is the
> usual frustration...
Ryan and me have been communicating on a hardware mod on rem - one of
the mipsel buildds. Basically the machine is running on a PIO based ATA
disk because of the fundamentally broken design of the IDE interface
(abusing some leftover GPIO pins). I already got hold of a DMA capable
PCI controller which will be put into the machine most likely end of
this week and will bring a performance boost as it has been seen on
other BCM based buildds.
So - the issue is known - its worked on - most likely not the speed
everyone is happy with ...
Flo
PS: There is also a kernel bug in the mips and mipsel kernels which got
triggered on the buildds lately which crashed them hard a couple of
times and caused delay by downtime - see debian-mips list.
--
Florian Lohoff flo@rfc822.org +49-171-2280134
Those who would give up a little freedom to get a little
security shall soon have neither - Benjamin Franklin
Attachment:
signature.asc
Description: Digital signature