[Zebralist] configure fails on dual xeon cluster

Wray Buntine buntine at hiit.fi
Mon Jan 1 16:07:35 CET 2007


Adam Dickmeiss wrote:

> Wray Buntine wrote:
>
>> Note I did test it under Zebra 2.0.6 sorry about the confusion.
>> See the comments below.
>>
>> marc wrote:
>>
>>> Wray Buntine wrote:
>>>
>>> Hi Wray
>>>
>>> what seems to happen is that our check for long file system (LFS) is 
>>> not working correct on your platform.
>>>
>>> If you have a look into configure.ac, you see that there is a small 
>>> C program to figure this out.
>>>
>>>
>>> Please try to save the following (cut'n paste from configure.ac) 
>>> into a file
>>> 'lfs.c' 
>>
>>
>> ...
>>
>>>
>>>
>>>
>>> and try to compile:
>>>
>>>  gcc lfs.c -o lfs
>>>
>>> then, try to run and check for exit value:
>>>
>>>  ./lfs || echo '64'
>>
>>
>> OK, so its 'lfs' that hangs.
>> I tried running it under gdb, and its fcnt() that hangs.
>> I'm willing to give you a temporary account on the machines if 
>> necessary.
>
>
> What file system are you using?

Bingo ... right you are, it was via NFS.
OK, switched to local disks (/tmp is all we have on clusters)
and everything works.

Wow, got my problem in two simple steps.
You guys at  IndexData are good!

Thanks,  Wray



More information about the Zebralist mailing list