This was brought up before, and I believe the problem was that for some
of the older openafs client/server interactions, this causes a major
hang, so that it just sat there for about 2-5 minutes.
I could double check this though and see if that still is the case.
Troy
Michel Jouvin wrote:
> Along the same lines, if there is any new AFS errata for SL 3.07, it'd
> be nice to replace the 'wget' command in init script by the rxdebug
> command used in openafs 1.4.1. 'wget' is a problem when the port is
> firewalled but the router/firewall doesn't answer a connection refused
> (to avoid DoS).
>
> Michel
>
> --On jeudi 28 septembre 2006 12:02 +0200 Michel Jouvin
> <[log in to unmask]> wrote:
>
>> Troy,
>>
>> I'd suggest, in the new release, to add a test in the startup script for
>> a file ThisCell with an empty line (currently only an empty file is
>> tested). Here is the proposed fix :
>>
>> --- /tmp/afs 2006-09-28 11:58:00.000000000 +0200
>> +++ /etc/init.d/afs 2006-09-28 11:58:40.000000000 +0200
>> @@ -83,6 +83,11 @@
>> return 1
>> fi
>> CELL="$(cat /usr/vice/etc/ThisCell)"
>> + if [ -z "$CELL" ]
>> + then
>> + echo afs: ThisCell is empty. Not starting. 1>&2
>> + return 1
>> + fi
>> grep -w -q -s ">$CELL" /usr/vice/etc/CellServDB
>> if [ $? -ne 0 ] ; then
>> echo afs: $CELL from ThisCell is not in CellServDB. Not
>> starting. 1>&2
>>
>>
>> Michel
>>
>> --On mercredi 27 septembre 2006 18:45 +0200 Michel Jouvin
>> <[log in to unmask]> wrote:
>>
>>> Troy,
>>>
>>> Thank you very much. It works !
>>>
>>> Michel
>>>
>>> --On mercredi 27 septembre 2006 09:28 -0500 Troy Dawson
>>> <[log in to unmask]>
>>> wrote:
>>>
>>>> Hi Michel,
>>>>
>>>> They are now in the testing area.
>>>> As I said earlier, when we release S.L. 4.4, we'll move the openafs
>>>> 1.4.1 packages into the errata. That hopefully will be towards the end
>>>> of next week.
>>>>
>>>> For now you can install them by hand from
>>>> ftp://ftp.scientificlinux.org/linux/scientific/40rolling/testing/x86_64
>>>> /R PMS/openafs
>>>>
>>>> rpm -Uvh
>>>> ftp://ftp.scientificlinux.org/linux/scientific/40rolling/testing/x86_64
>>>> /R
>>>> PMS/openafs/kernel-module-openafs-2.6.9-34.ELlargesmp-1.4.1-0.11.SL.x86
>>>> _6 4.rpm
>>>>
>>>> or by yum
>>>>
>>>> yum --enablerepo=sl-testing install
>>>> kernel-module-openafs-2.6.9-34.ELlargesmp
>>>>
>>>> Troy
>>>>
>>>> Michel Jouvin wrote:
>>>>> Currently I am still running the original SL 4.3 kernel, 2.6.9-34.
>>>>>
>>>>> Thanks.
>>>>>
>>>>> Michel
>>>>>
>>>>> --On mercredi 27 septembre 2006 08:39 -0500 Troy Dawson
>>>>> <[log in to unmask]> wrote:
>>>>>
>>>>>> Michel Jouvin wrote:
>>>>>>> Is a RPM available to provide kernel-module-openafs for largesmp
>>>>>>> kernels in SL 4.3 (i386 and/or x86_64) ? largesmp kernel seems to be
>>>>>>> installed by default on (Opteron?) SMP machines.
>>>>>>>
>>>>>>> Thanks in advance for any help.
>>>>>>>
>>>>>>> Michel
>>>>>>>
>>>>>>
>>>>>> Hi,
>>>>>> I just tested my build scripts. It seems that largesmp for x86_64
>>>>>> and
>>>>>> hugemem for i386 don't build with our openafs 1.4.0 rpm, but do build
>>>>>> with out openafs 1.4.1 rpm.
>>>>>> We haven't pushed openafs 1.4.1 out to the errata yet, because I was
>>>>>> waiting until we released S.L. 4.4 ... just incase we had any last
>>>>>> minute changes to it.
>>>>>> openafs 1.4.1 is in our testing area though, and I'll put the kernel
>>>>>> modules I just built up there.
>>>>>> Is there any particular kernel you need, or just the latest one?
>>>>>> (When we release openafs as an errata, we'll have all the kernels,
>>>>>> but I currently haven't built them all yet.)
>>>>>>
>>>>>> Troy
--
__________________________________________________
Troy Dawson [log in to unmask] (630)840-6468
Fermilab ComputingDivision/CSS CSI Group
__________________________________________________
|
