The problem here is that people are charging forward with a solution
without discussing (or even considering) all the issues involved.
Thanks for starting this thread so we can do that.
As I understand it, the problem at hand is:
1) Apache chokes for all users if the path to any single user's
logfile does not exist
2) We want to put a limit on the amount of disk any single logfile
can consume
If this is the problem we're trying to solve, there are far simpler
ways to do it than creating O(numusers) more volumes. There are
plenty of log rotators out there that handle #2 "out of the box" and
which can also be set to discard logs if the logfile is missing
rather than causing problems.
I've seen references on and off to what appears to be a third problem,
which is almost orthogonal to the previous two issues:
3) For some reason we don't want users to be able to delete or modify
their own logs.
Is this actually a goal? I don't care about the policy angle, but
technically it opens a whole new can of worms. Apache runs with the
user's tokens -- how is it going to write to the logfile if the user
can't write to it?
> Now megacz has been making recommendations which seem to be based on
> viewing volume creation and/or maintenance as a significant cost,
That's something of an oversimplification; there's always a tradeoff
here in terms of management complexity. The tradeoff is often obvious
for one-off volumes (common.debian, common.etc) -- "just create more
volumes". It is less obvious for volumes which get multiplied by the
number of users.
- a
_______________________________________________
HCoop-SysAdmin mailing list
[email protected]
http://hcoop.net/cgi-bin/mailman/listinfo/hcoop-sysadmin
