> This is known as the "ls of death". If a user should mistakenly do a 
> "ls /afs" when the afs servers use a wide number sites, linux appears to 
> hang. Response times from MacMurdo station (antartica) really bite. I 
> think they use "tcp/ip over penguin". 

While adding better signal handling would fix this particular problem,
you'd still lose when, say, a file manager tried to open up /afs, since
you can't easily and safely interrupt it with ^C.  I've been meaning to
implement fakestat -- i.e., create fake information for volume roots
without actually talking to the vldb or fileservers.  This would fix
the problem for both command-line and graphical tools, on all platforms.
(Now I just need to actually do it.. :-)

-- kolya
_______________________________________________
OpenAFS-devel mailing list
[EMAIL PROTECTED]
https://lists.openafs.org/mailman/listinfo/openafs-devel

Reply via email to