Odd problem with a change not propagating..

[mmarion@qualcomm.com]

On  4 Sep, Tomas Olsson wrote:

> We had some reports about an invalidation problem when setting a new
> sysname, could be the same thing. It didn't happen whenn I tried, and left

Sounds like it might be the same.. though change from no sysname to a link
with sysname.

> it at that for the time being. Setting a new sysname isn't a very common
> operation. I'll take a look at it when I find the time.

Right.. we don't do it very often either.

> Good thing to know. If you happen to find a simple test case, please tell us.

Will do.

> Apart from this, is everything working well? What's you setup, usage, etc?

Yes.. working fine.  Using arla so far only on amd64_linux26 hosts (SLES9).
No problems reported yet.  Much better then openafs was working for us in
64bit hosts.  I ran into problems on ia64 and am64_linux24 with openafs where
any usage of existing cache files (after restarting or rebooting) would result
in corruption, like symlinks suddenly pointing to nothing, or pointing to 8bit
ascii characters.  I finally worked around this by making our init script blow
away the existing cache everytime.    Arla has been fine using the existing
cache.

-- 
Mike Marion-Unix SysAdmin/Staff Engineer-http://www.qualcomm.com
It took the power of 3 Commodore 64's to go to the moon, but it takes a 
486 to run Windows... Something is desperately wrong here. - Stolen from /.