Re: 2.0.31p2 locks up

Stephen R. van den Berg (srb@cuci.nl)
Tue, 5 Aug 1997 23:34:15 +0200


Andre Uratsuka Manoel wrote:
>On Wed, 30 Jul 1997, Alan Cox wrote:
>-> > After about a day the number of skbufs is about 100 000. What is
>-> > the size of a skbuf? If they averaged 100 bytes, I would have 10 MB
>-> > taken by kernel memory.

>-> What network devices are you using. Do you have any unusual routes, are
>-> you trying to use the traffic shaper with 2.0.30 ?

> More unusual things: alias are assigned to different networks as
>eth0. Shaper patch was applied, but the shaper module is not loaded in.
>Also, the shaper patch was not applied in one of the machines with
>problems.

Aha! I think we might have struck gold here.
What I have:
- Several machines humming along smoothly (apparently).
- One machine getting real low on memory every now and then (the watchdog
eventually reboots it).
All machines run 2.0.30p2+werner-fink-the-1st(with the old buffer code)

The only real difference between this machine and all the others are the
numerous (20 or so) IP-aliases it carries which are *not* in the network
attached to eth0.

It runs firewalling, ppp-logins, gated (doing OSPF multicasts), and virtual
webservers.
It does *not* run transparent proxies, traffic shapers, always-defragment.

-- 
Sincerely,                                                          srb@cuci.nl
           Stephen R. van den Berg (AKA BuGless).
Sign at a nursery school: Please pick up your children promptly after service.
                          No refunds after Wednesday.