Red Hat Bugzilla – Bug 56721
K12LTSP: bump loop detection to 250/s
Last modified: 2007-04-18 12:38:26 EDT
Please bump the loop detection to 250/sec in order to
support PXE boots from thin clients:
Here's the change:
> * Wed Oct 24 2001 Eric Harrison <email@example.com>
> - LTSP HACK: bumped up loop detection to 250/second: PXE booting
> can be pretty hard on xinetd.
diff xinetd-7.2/xinetd.sysconf xinetd/xinetd.sysconf
> EXTRAOPTIONS="-loop 250"
Uh... why? It's a configuration option. It's not impossible to do afterwards,
and if implemented would limit the usefulness of loop detections in the first
place. A special case for use often needs a little configuration.
Um, please remember I'm just the messanger here.
Because thin clients have different needs, and, particularly in
the case of k12ltsp, removing some of the deeper darker
configuration voo-doo is gonna be neccessary.
Please feel free to suggest a better number than 250 if you wish.
However, understand that the number was arrived at with
Real World PXE booting experience, something we are sorely
lacking in at the moment.
The loop detection in 2.3.3 and earlier is clumsy, and duplicates some of the
functionality in the configuration section. It's clumsy because the -loop
option is passed on the command line and can't be overridden in the config
The current development branch of xinetd removed the -loop option, since the
per-service 'cps' directive (connections per second) duplicated much of the
purpose of -loop, but is more configurable. It still defaults to on, with the
old compiled defaults, but is at least configurable now.
The current rawhide version of xinetd is based on that snapshot. Reassigning to
pxe - if there is a desire to change the defaults for that package, it can be done.
Leaving defaults as-is