lkml.org 
[lkml]   [1999]   [Jan]   [25]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
Date
Subject2.2.0-pre9 blow up

Hi. Just had a strange thing happen to my main machine today.
It had been happily running 2.2.0-pre9, and then for some unknown
reason I noticed that the load was climbing into the sky...

Lots and lots of things were getting into " D " (disk wait) state, and
not coming back. It looked like anything that tried to exit, was just
not doing so. I quit my netscape and it just sat there in "D" and
wouldn't free up it's memory.

I tried a 'swapoff -a' and then 'swapon -a', but that didn't change
anything. (It did work tho...swapped everyhing back in ok)

At this point the load was up to about 100 or so. The machine was
still very responsive, it was just all the disk waiting that was
making the load climb. Anything that did an exec was likely to get
stuck in disk land, I tried a 'sync' in one window and it never came
back. I had to use alt-sysrq-b to reboot.

I saved off a demsg output (that had some interesting stuff in it),
but after I rebooted, all the files I had made got nuked by fsck. ;(

I did copy down by hand some alt+sysrq output:

alt-sysrq-p:

EIP: 0010:[<c0107347>] EFLAGS: 00000246
EAX: 0000001f EBX: c03d6000 ECX: c03d6000 EDX: c03d6000
ESI: c023sf78 EDI: c0106000 EBP: 00001000 DS: 0018 ES: 0018
CR0: 8005003b CR2: 4000b000 CR3: 00101000

alt-sysrq-t:

tons of output...a sample:

HTTPD: 221 D c023400 0 15176 344 15179 15175
sig: 0 0000000000000000 0000000000000000 : X

alt-sys-rq m:

mem-info
free pages: 2184kb
(free 546 (256 512 768)
152*4kb 45*8kb 12*16kb 2*32kb 1*64kb 7*128kb = 2184kb)
swap cache: add 83898, delete 83818, find 546839/667901
free swap: 122568kb
32768 pages of RAM
834 reserved pages
11561 pages shared
80 pages swap cached
0 pages in page table cache
buffer memory: 2620kb
buffer heads: 2656
buffer blocks: 2620
clean: 2481 buffersm 8 used (last=2401) 0 locked, 0 protected, 0 dirty
locked: 138 buffers, 85 used (last=137) 0 locked, 0 protected, 0 dirty
dirty: 1 buffers, 0 used (last=0)
networking buffers in use: 597
total network buffer allocations: 68775817
total failed network buffer allocs: 0
tp fragment buffer size: 0

alt-sysrq-s didn't appear to sync the disks. I didn't hear any
activity, and filesystems came up dirty. alt-sysrq-u also didn't
appear to do anything.

I saved off a dmesg and some ps listings, and then rebooted.
On bootup fsck nuked all the files I had created, so I don't have that
data anymore. ;(

syslogd was one of the first to get locked in D state, so nothing in
the logs. ;(

Hopefully this is helpfull to someone. I can lookup stuff in my system
map if anyone thinks it's usefull to do so. Perhaps something not
giving up a filesystem lock?

I am running 2.2.0-pre9 again, so if this happens again, what usefull
info can I get?

kevin

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:50    [W:1.194 / U:0.252 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site