On 18/10/06 at 11:13 +0200, Filippo Giunchedi wrote: > On Wed, Oct 18, 2006 at 09:36:26AM +0200, Lucas Nussbaum wrote: > > > Nice. How much time would a complete archive rebuild take? > > > > The build of the 9716 source packages which I was able to build on AMD64 > > took a total time of 790936s (9 days, 4 hours). So it should > > theoritically be possible to build in a bit more than 2 hours, but some > > packages seem to take too long for that: > > will the grid improve performance by using tools like ccache? This might > improve > rebuild times though, not first-time builds.
I could use it, yes. But it would require a shared storage somewhere, and I fear that the network/storage pressure will be too important. It's probably easier for me to just increase the number of nodes if I want to build faster. (but this doesn't solve the "slow builds" problem). > Anyway, how is the build already splitted across nodes automatically? or > something like make -j<largeinteger> should do? (this is only out of > curiosity) I'm keeping it as simple as possible. A special node is in charge of distributing packages amongst nodes. It just ssh to the node and run a wrapper around sbuild. So each build happens on only one node. Nodes are bi-amd64, but most builds probably only use one CPU. I'm not really interested in building faster currently: I would prefer to build more reliably (have less failures). I fear that increasing complexity of the build environment will also increase the number of false positives. -- | Lucas Nussbaum | [EMAIL PROTECTED] http://www.lucas-nussbaum.net/ | | jabber: [EMAIL PROTECTED] GPG: 1024D/023B3F4F | -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]