>>> [EMAIL PROTECTED] said...
> If you have a very large number of users, the central AFS-based spool
> directory is probably not a good idea. First, it places too many things
> in one directory, causing quota and access problems. It would be better
> to have a subdirectory for each user, which will speed access and
> allow separate ACLs for each user. Quotas will still be a problem.
> If the mail spool directory is on one fileserver, no mail can be
> delivered while that machine is out of service, which has been a
> serious consideration for my cell.
How much is the performance degradation when you use AFS-based
central spool directory?
> > We encounters some problems such as file locking, ACL setting,
> > MUA recompiling, user tranining, etc.
>
> I advise using only locking by lock file, since neither flock nor
> lockf will work across all platforms. With respect to ACLs our solution
> requires system:administrators or a mail group to have all access,
> and users themselves to have write access at least. Then the reading
> programs (like elm) do not need to know about ACLs. I have had problems
> with users taking away the mail write access from their directories,
> which simply prevents them from receiving mail. No special user
> training should be necessary.
What we think as a problem is that many users don't want their home
directory to be touched and so many users don't know well about ACL's and
mail delivery mechanism. Also many users can accidentally screw up their
home directory.
So, we'd prefer the central mail spool if the performance is not
bad. Would you tell me how bad the performance is when you set up this
central mail spool? We have about 2000 users currently.
--
___________________________________________________________________________
_____ ___ __ __
__/\__ |_ _| / _ \ \ \_/ / Remember Yesterday, Dream about Tomorrow
\ / | | | /_\ | \ / but ... LIVE TODAY !!!
/_ _\ | | | _ | | |
\/ __| | | | | | | | [EMAIL PROTECTED] (Jae-young Kim)
|___/ |_| |_| |_| Researcher, Computer Center/POSTECH, KOREA
___________________________________________________________________________
