| 1 |
Release Critical: |
| 2 |
* chmod TMPDIR/native, as otherwise set umask prevent it from being readable by the client |
| 3 |
I'm wondering if it doesn't make sense to leave the daemon out of the game and let the |
| 4 |
client package the environment - it could either cache it in /tmp and leave a note in |
| 5 |
~/.icecream or it packages it right into ~/.icecream and use e.g. the inode numbers of |
| 6 |
the compilers to verify it's the right environment - think NFS users changing machines. |
| 7 |
This would simplify some things and avoid the above bugs (and it would make it more |
| 8 |
convenient for users of /opt/gcc-cvs/bin too) |
| 9 |
This would also help users who are running in a chroot enviornment where the local |
| 10 |
daemon cannot even find the path to the compiler |
| 11 |
|
| 12 |
* Improve Documentation (cschum) |
| 13 |
* make the protocol version an uint32, not a hand-build array. |
| 14 |
* let the client specify it was compiled with recent compiler (supporting -param). If so, |
| 15 |
let the daemon compile with the options, otherwise leave them out. |
| 16 |
|
| 17 |
Akademy must have's: |
| 18 |
* daemon: only accept connections from a sane net prefix (scheduler can determine |
| 19 |
this and send it out via confMsg) to avoid being exposed to the internet |
| 20 |
* benchmark/testing jobs (broken machines with broken ram and broken network card on the |
| 21 |
icecream network destroy the fun and are not easy to find manually) |
| 22 |
|
| 23 |
Random: |
| 24 |
* Option -iiface to specify the interface[s] to listen on for the scheduler, or |
| 25 |
to use for the daemon. |
| 26 |
* Don't explicitly forbid tunnels because it could be useful for things like |
| 27 |
vmware |
| 28 |
* If someone calls a amd64 client on a host that runs a ia32 daemon and there are |
| 29 |
no other amd64 daemons in the farm, he will get no answer, but a timeout from |
| 30 |
scheduler (quite a corner case, but neat) |
| 31 |
* use syslog |
| 32 |
* Log problems found at some scheduler log - or even in the monitor. E.g. if a client |
| 33 |
can't reach a given daemon, it should be able to tell. Perhaps the scheduler can even |
| 34 |
disable that very host for some penalty time |
| 35 |
* Reduce amount of force-waits, especially if they involve network latency (scheduler queries) |
| 36 |
and daemon context switches: |
| 37 |
- remove the need for EndMsg |
| 38 |
- do not ask the daemon about the first job (WIP dirk) |
| 39 |
* if a compile job SIGSEGV's or SIGABORTs, make sure to recompile locally because it could |
| 40 |
be just a glibc/kernel incompatibility on the remote site |
| 41 |
* Add heuristic which optimises for overhead-reduction. Statistics prove ( ;) ), that for |
| 42 |
e.g. linux kernel the file size varies a lot, and small jobs should be preferably compiled |
| 43 |
locally and bigger ones preferably remote. |
| 44 |
* Split number of jobs into number of compile jobs (cheap) and number of non compile jobs (can |
| 45 |
be expensive, e.g. ld or meinproc). The reason is that multicore chips become more and more |
| 46 |
common. Today you can get quad cores easily and in some month we've 8 cores and several link |
| 47 |
jobs at the same time can pull down your whole system. |
| 48 |
(I can life with a hard coded number of one non compile job but others probably prefer a |
| 49 |
configurable number) |
| 50 |
* Consider launching a scheduler on-demand if there is none available or if a daemon knows |
| 51 |
it has a better version than the scheduler that is available (https://github.com/icecc/icecream/issues/84). |
| 52 |
|
| 53 |
Suggestions from "Wilson Snyder" sicortex.com: |
| 54 |
- Add ICECREAM_RUN_ICECCD to make the scheduler machine not run iceccd |
| 55 |
|
| 56 |
Christoph Thielecke writes: |
| 57 |
> 1) Problem von icecream: icecc hat /usr/bin/gcc-4.2 kopiert (ist aber ein |
| 58 |
> Symlink auf den Hardening-Wrapper, der installiert war, siehe Listing |
| 59 |
> unten). Kannst Du vieleicht eine Prüfung einbauen, ob gcc-x.y mit .real |
| 60 |
> endet bzw der Wrapper ist? |
| 61 |
|
| 62 |
Andi Kleen writes to protect against misconfiguration: |
| 63 |
>generate somefile.h on the client with |
| 64 |
>#if __GNUC__ != expected_major || __GNU_MINOR__ != expected_minor |
| 65 |
>#warning blabla |
| 66 |
>#endif |
| 67 |
>and then run with |
| 68 |
>-include somefile.h |
| 69 |
> (this file needs to be in the environment created by create_env) |
| 70 |
|
| 71 |
From thedrow (distcc maintainer) |
| 72 |
>Merging this back might just mean feature parity with what icecream currently |
| 73 |
>provides or merging some of the code directly to distcc. |
| 74 |
>I think that the entire Linux community will gain from one tool that is able to |
| 75 |
>distribute compilation of large C/C++ components that provides more flexibility |
| 76 |
>and allows you to grow from the easy to setup distcc type of deployment to the 100 |
| 77 |
>machines build farm that icecc provides. |
| 78 |
>Are there more features that diverged from distcc except for the server/client |
| 79 |
>architecture? |
| 80 |
We think the the only other feature is better support for automatic chroot. |
| 81 |
|