Date: Tue, 19 Sep 2000 20:29:18 +0200 From: Tor.Egge@fast.no To: root@nihil.plaut.de Cc: grog@lemis.com, current@FreeBSD.ORG, freebsd-smp@FreeBSD.ORG Subject: Re: Debugging -current SMPNG HANG on heavy disk-io Message-ID: <200009191829.UAA68967@midten.fast.no> In-Reply-To: Your message of "Sun, 17 Sep 2000 16:29:41 %2B0200 (CEST)" References: <Pine.BSF.4.21.0009171622240.840-100000@localhost>
next in thread | previous in thread | raw e-mail | index | archive | help
> (kgdb) ps
> pid proc addr uid pri ppid pgrp flag stat comm wchan
> 37 c7874a00 c9665000 0 32 6 36 004086 3 tar piperd c9663f20
> 36 c7874bc0 c960a000 0 32 6 36 004006 3 tar FFS node c02f4220
This looks like you've hit the limit for the FFS node memory type.
vmstat -m will indicate if this is correct.
If you see somethinig like
Memory statistics by type Type Kern
Type InUse MemUse HighUse Limit Requests Limit Limit Size(s)
[....]
FFS node262144 65536K 65536K 65536K 2024460 0 6 256
[....]
Memory Totals: In Use Free Requests
93897K 608K 9482590
(i.e. MemUse == Limit), then you've hit the limit. The process
allocating a FFS node normally holds a vnode lock, resulting in
a cascade of vnode locks and a frozen system.
Increasing the kmem_map size (by setting a loader variable
(kern.vm.kmem.size) or defining VM_KMEM_SIZE and VM_KMEM_SIZE_MAX in
the kernel config file) should help.
- Tor Egge
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-smp" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200009191829.UAA68967>