Re: Debugging kernel memory allocation

To: Tom Ivar Helbekkmo <tih%hamartun.priv.no@localhost>

Subject: Re: Debugging kernel memory allocation

From: Rich Neswold <rich.neswold%gmail.com@localhost>

Date: Fri, 21 Nov 2014 11:33:17 -0600

On Mon, Nov 17, 2014 at 2:42 AM, Tom Ivar Helbekkmo <tih%hamartun.priv.no@localhost> wrote:

Lars Heidieker <lars%heidieker.de@localhost> writes:

> are any pools with a limit are exhausted?

Here's a possibly interesting data point:

Here's my interesting data point(s):

I have a Supermicro X7DCL (8-core, SATA drives, 24GB RAM), a Dell 2850 (8-core, SCSI drives, 6GB RAM) and another Dell 2850 (4-core, SCSI drives, 6GB RAM.) Both the 8-core machines are being used, so I shouldn't try to trigger the stalls on them. I've been trying to reproduce them on the unused 4-core system, though. I've built the world twice and traversed the PKGSRC tree many times and I can't seem to get it to happen. Today, however, I did a 'ls -Rl' on the 8-core Dell on a relatively small tree (much smaller than PKSRC) and the system stalled for 40 seconds.

So, right now, I've only seen the stalls on my two 8-core machines. I haven't been able to stall the 4-core machine. If I can schedule some downtime, I'm thinking of moving the services to the 4-core system. Then I'll have the 8-core Dell to try to debug. Unfortunately, I'm not there yet.

Rich