Bug 30283

Summary:	swap_pager problem with 4.2-RELEASE
Product:	Base System	Reporter:	david <david>
Component:	kern	Assignee:	freebsd-bugs (Nobody) <bugs>
Status:	Closed FIXED
Severity:	Affects Only Me
Priority:	Normal
Version:	4.2-RELEASE
Hardware:	Any
OS:	Any

Description david 2001-09-03 11:30:06 UTC

After approx 90 days uptime one of our mail servers running 4.2-RELEASE failed with the following errors continuously repeating on the console:

swap_pager: indefinite wait buffer: device: #ad/0x20001, blkno: 18920, size: 4096
swap_pager: indefinite wait buffer: device: #ad/0x20009, blkno: 27088, size: 4096

Although the server still responded to pings etc, any disk activity would cause the respective process to hang.

The swap configuration is as follows:

Device          1K-blocks     Used    Avail Capacity  Type
/dev/ad0s1b        524160    11536   512624     2%    Interleaved
/dev/ad1s1b        524160    11564   512596     2%    Interleaved
Total             1048320    23100  1025220     2%

Fix: 

Unknown
How-To-Repeat: 
Unknown

Comment 1 Peter Pentchev 2001-09-03 11:38:18 UTC

On Mon, Sep 03, 2001 at 03:27:51AM -0700, David Hedley wrote:
> 
> >Number:         30283
> >Category:       kern
> >Synopsis:       swap_pager problem with 4.2-RELEASE
> >Originator:     David Hedley
> >Release:        FreeBSD 4.2-RELEASE
> >Organization:
> Intelligent Network Technology Ltd
> >Environment:
> FreeBSD freddy.inty.net 4.2-RELEASE FreeBSD 4.2-RELEASE #0: Fri Dec 15 15:51:22 GMT 2000     david@freddy.inty.net:/usr/src/sys/compile/GENERIC+MODS  i386
> 
> >Description:
> 
> After approx 90 days uptime one of our mail servers running 4.2-RELEASE failed with the following errors continuously repeating on the console:
> 
> swap_pager: indefinite wait buffer: device: #ad/0x20001, blkno: 18920, size: 4096
> swap_pager: indefinite wait buffer: device: #ad/0x20009, blkno: 27088, size: 4096
> 
> Although the server still responded to pings etc, any disk activity would cause the respective process to hang.
> 
> The swap configuration is as follows:
> 
> Device          1K-blocks     Used    Avail Capacity  Type
> /dev/ad0s1b        524160    11536   512624     2%    Interleaved
> /dev/ad1s1b        524160    11564   512596     2%    Interleaved
> Total             1048320    23100  1025220     2%

AFAICS from a quick look at /sys/vm/swap_pager.c, this error message
is caused by a timeout on the actual read operation from the swap
device.  I think you should check your disks for physical errors,
or better just replace them, before damage is done to more than
the swap partitions..

And then, of course, I might be totally off-base; some VM/swap guru
to explain what exactly does the 'indefinite wait buffer' message mean?

G'luck,
Peter

-- 
This sentence every third, but it still comprehensible.

Comment 2 iedowse freebsd_committer

2002-11-14 01:40:23 UTC

State Changed
From-To: open->feedback


Does this problem still occur?

Comment 3 iedowse 2002-12-01 00:45:17 UTC

Adding to the audit trail:

In message <JPEJJOLCCBGIGJGODCGGIEKKCGAA.david@inty.net>, David Hedley writes:
>
>Don't know - it only occurred when there were two heavily used disks on =
>the same IDE bus.
>
>We replaced the two disks with just one bigger disk and we haven't seen =
>the problem since.
>
>Haven't tried a later FreeBSD build on it yet.
>
>David

Comment 4 iedowse freebsd_committer

2002-12-01 00:53:37 UTC

State Changed
From-To: feedback->closed


There is insufficient information available to determine if the 
effect was caused by a FreeBSD bug or a hardware issue. As the 
problem has not repeated itself, there is little use in keeping 
this PR around, but it can be re-opened if the issue reappears.