=?gb2312?B?tPC4tDogtPC4tDogtPC4tDogW2Rpc3RjY10gUHJvYmxlbSBhYm91dA==?= =?gb2312?B?IERpc3RjYyBvbiBXaW5kb3dzIG5vdCBjeWd3aW4gdG8gY29tcGlsZSBBUk0=?= =?gb2312?B?IHByb2plY3Q=?=

[Feng Tan]

That makes some sense, and this is no ADS installed on 192.168.0.232.

 

But I still have a question about the 

 

> Distcc [5364] (dcc_check_backoff) still in backoff period for
revo-lufei/4,lzo

 

Why did this happen and how to avoid it? Sorry that I can¡¯t get answer from
the debug information

  _____  

·¢¼þÈË: Fergus Henderson [mailto:fergus at google.com] 
·¢ËÍʱ¼ä: 2008Äê10ÔÂ7ÈÕ 21:34
ÊÕ¼þÈË: Feng Tan
³­ËÍ: distcc at lists.samba.org
Ö÷Ìâ: Re: ´ð¸´: ´ð¸´: [distcc] Problem about Distcc on Windows not cygwin to
compile ARM project

 

> distcc[5364] ERROR: compile verno/verno.c on 192.168.0.232/8,lzo failed
with exit code 110

 

exit code 110 is "compiler not found".

Check that you have "tcc" installed on that host (192.168.0.232), and that
it's in the distcc server's PATH.

(Or try using an absolute path to "tcc".)

Also check the server log file on that host; that may have more clues.

 

> distcc[5364] (dcc_check_backoff) still in backoff period for revo-lufei/4,
lzo

 

This occurs because there was previous an error for that host, in the last
minute or so.

distcc won't keep retrying on hosts where it has previously failed.

(rm -rf ~/.distcc/

If all of your distcc servers other than one have problems with "tcc" not
being found,

then that would explain why you're only getting parallelism suitable for a
single distcc server.

 

2008/10/7 Feng Tan <feng.tan at revoview.com>

Ok, I attached the whole debug file and it's a little bigger. 

And in this file I closed pump mode.

 

As for the pump mode, I have tried to close this, and got the same result.

 

So I don't think the pump mode is the problem.

  _____  

·¢¼þÈË: Fergus Henderson [mailto:fergus at google.com] 

·¢ËÍʱ¼ä: 2008Äê10ÔÂ7ÈÕ 16:08

ÊÕ¼þÈË: Feng Tan
³­ËÍ: distcc at lists.samba.org

Ö÷Ìâ: Re: ´ð¸´: [distcc] Problem about Distcc on Windows not cygwin to
compile ARM project

 

2008/10/7 Feng Tan <feng.tan at revoview.com>

Thanks Fergus Henderson.

The make version is as below:

GNU Make 3.81

This program built for i686-pc-mingw32

DISTCC_HOSTS is set as windows environment variables and every item is like
"feng/4,cpp,lzo" .

 

 

And I attached the out.txt file as your reference.

The out.txt file looks incomplete... it shows only the first half of a
couple of compilations.

We'd probably need the full log from a "make" run.

 

But one thing sticks out:

 
distcc[5680] Warning: INCLUDE_SERVER_PORT not set - did you forget to run
under 'pump'?

PS: I have tried pump mode but failed, which should not influence much 

If you're not using the pump script, then distcc won't use pump mode, and
then you'll be limited by the preprocessing speed on the local machine.

If the remote machines can can compile faster than the local machine can
preprocess, that could explain why you're never getting more than about 4
concurrent jobs.  Of course there are other possible explanations too, e.g.
maybe something is broken with the locking code.

 

-- 

Fergus Henderson <fergus at google.com>




-- 
Fergus Henderson <fergus at google.com>

-------------- next part --------------
HTML attachment scrubbed and removed