Date: Thu, 09 Jul 1998 13:56:11 -0700 From: Julian Elischer <julian@whistle.com> To: current@FreeBSD.ORG Subject: TESTERS NEEDED: Softupdates looks Very good. Message-ID: <35A52E6B.6201DD56@whistle.com>
next in thread | raw e-mail | index | archive | help
As of this morning there are 2 known bugs in Softupdates,
and one known 'gotcha'
The bugs are:
1/
the following "rare" panic has not been traced yet as we have been
unable to get a core file with it..
panic: newdirrem: inum 26368 should be 26367
2/
there is also a dependency order problem in some cases
(very rare)
Over all a crash with soft updates will already probably leave your
disk in a much cleaner state than you're used to.
The "gotcha" is really quite hillarious..
This note from Matt Dillon at Best Communications..
------ Begin quote------
unlink() is so fast under softupdates that Diablo's news spool
expiration removed 80% of the files in the test news spool before
it realized that it had sufficient free space. Oops!
------
This is because the unlink happens asynchronourly, and the
space may not show up for 30 seconds or so.
Unlink() is now 2 orders of magnitude faster..
in 30 seconds you could delete several million files,
(and it apparently did). Unlinking the entire X11R6 hierarchy
takes well under 1 second. It's now possible to type
#rm -rf /
and REALLY screw yourself before you have time to hit ^C :-)
We need more testers. specifically to try get a core dump of that panic.
If you know how to compile a debug kernel
config -g MYKERNEL
cd ../../compile/MYKERNEL
make depend; make
cp kernel kernel.debug
strip -d kernel
make install
and you know how to get a core dump
dumpon /dev/rsd0b
and you know how to run softupdates
cd /sys/ufs/ffs
cat *READ*
then YOU are needed for testing!
if you get a core-dump and you have a matching debug kernel,
let us know asap.
julian
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?35A52E6B.6201DD56>