On Tue, 14 Aug 2007, Derrick J Brashear wrote:
Per our private thread, the salient point is "on linux" and not simply 1.2
versus 1.4
Yes, sorry, I'm guilty of not checking a 1.4.4 client on Solaris before -
which behaves correctly. I also have to add that my linux build is not
pristine 1.4.4 but has a number of patches applied I pulled from CVS.
I'll check 1.4.4 as released, but this will have to wait until tomorrow.
At least I know it's not intentional now.
Thanks for caring,
Stephan
On Tue, 14 Aug 2007, [EMAIL PROTECTED] wrote:
The AFS client used to cache writes on the local disk until a file is
closed, but this seems to have changed?
Consider the following script:
#!/bin/sh
while [ 1 = 1 ]; do
date
sleep 10
done
Now I run "script.sh >> afs12.out" on a 1.2.13 client, and "script.sh >>
afs14.out" on a 1.4.4 one. Except for the AFS client version, the systems
are identical (Scientific Linux 3, same kernel & patches, ...).
On any 3rd client system (version doesn't matter), I see afs14.out growing
by 30 bytes every 10 seconds, while afs12.out stays at size 0 until I kill
the script.
Is this change intentional? I always thought the old behaviour really helps
in environments with many clients? Any insights much appreciated.
Thanks,
Stephan
--
Stephan Wiesand
DESY - DV -
Platanenallee 6
15738 Zeuthen, Germany
_______________________________________________
OpenAFS-info mailing list
[email protected]
https://lists.openafs.org/mailman/listinfo/openafs-info
_______________________________________________
OpenAFS-info mailing list
[email protected]
https://lists.openafs.org/mailman/listinfo/openafs-info
--
Stephan Wiesand
DESY - DV -
Platanenallee 6
15738 Zeuthen, Germany
_______________________________________________
OpenAFS-info mailing list
[email protected]
https://lists.openafs.org/mailman/listinfo/openafs-info
