I'm having excellent stability with linux 2.4.0-test11. My machine now
survives heavy nightly Amanda backups without being deadlocked on the
morning (the machine is a backup client only, and backups run every 2 days).
Here's my uprecords time:
# Uptime | System Boot up
----------------------------+-------------------------------------------------
1 20 days, 01:35:37 | Linux 2.4.0-test11 Sat Nov 18 18:06:11 2000
2 14 days, 23:22:14 | Linux 2.4.0-test7 Wed Oct 11 07:05:16 2000
3 13 days, 11:43:50 | Linux 2.4.0-test10 Tue Oct 31 12:38:06 2000
4 13 days, 11:31:53 | Linux 2.4.0-test7 Mon Sep 25 20:09:01 2000
5 9 days, 08:19:12 | Linux 2.4.0-test5 Sat Aug 19 14:26:14 2000
6 9 days, 04:46:42 | Linux 2.4.0-test7 Mon Sep 11 07:17:51 2000
7 8 days, 17:31:57 | Linux 2.4.0-test7 Sat Sep 2 10:27:05 2000
8 4 days, 11:22:46 | Linux 2.2.15 Sun Jul 2 23:24:40 2000
9 4 days, 03:36:24 | Linux 2.2.15 Tue Aug 1 08:52:56 2000
10 3 days, 19:49:11 | Linux 2.4.0-test11 Sat Dec 9 11:37:38 2000
----------------------------+-------------------------------------------------
-> now 3 days, 04:43:42 | Linux 2.4.0-test11 Wed Dec 13 07:48:25 2000
----------------------------+-------------------------------------------------
1up in 0 days, 15:05:29 | Sun Dec 17 03:37:28 2000
no1 in 16 days, 20:51:55 | Tue Jan 2 09:23:54 2001
The topline was with 2.4.0-test11-pre4, which did not lockup, but I had
to reboot it due to the presence of an NFS bug:
kernel BUG at /usr/src/linux-2.4.0-test11-pre4/include/linux/nfs_fs.h:164!
I now run 2.4.0-test11 final.
I had BAD experience with the linux-2.4.0-test12-pre4 and -pre6 kernels:
one locked up after 3 hours, and the other did not survive a full Amanda
backup.
My system disk is SCSI, and I have an extra IDE disk that I use for storage
only, on the hda interface (not using HPT). I have flashed the official RU
BIOS, and have 2 Celerons 466 Mhz, not overclocked.
I stress the machine rather heavily: I'm constantly running 2 rc5des clients,
VMWare, plus xmms. I do get some APIC errors, but far less than with
other kernels I've used.
There seems to be a combination of things making the -test11 series rather
stable on my BP6 (which is a rev 1.0 board with 300W PS). I will try another
test series if the current kernel crashes before reaching #1, otherwise I
will keep using this kernel until the official 2.4.0.
Raphael
--
=- To unsubscribe, email [EMAIL PROTECTED] with the -=
=- body of "unsubscribe linux-abit". -=
