Re: Kernel 2.0.30 pre-X SMP (in)stability report

Jon Tombs (jon@gte.esi.us.es)
Fri, 19 Sep 1997 00:53:49 +0200 (MET DST)

Thomas Schenk said:
>
>
> On 18-Sep-97 Agus Budy Wuysang wrote:
> >Before the real 2.0.31 is released, I would like to share
> >my own experience with the prepatches. So far I have found
> >that:
> >- Prepatch 2 is the most reliable/stable.
> >- Prepatch 5 & 9 is not stable for my hardware.
>
> I have not found this to be true. I have found that the latest prepatch (9)
> is quite stable on our hardware.

I have found that 2.0.31-pre9 is still unsuitable for SMP. On our Dual
Pentium 200 it will spontaneously reboot under the same conditions that
previously caused a deadlock (basically running wwwstats on several
10MB files of apache httpd logs).

No logs and no errors so I haven't got very far tracing the problem (and
I can't take the server out of service for long).

I suspect that the kernel does still deadlock in kernel mode, but the
anti-deadlock patches allow jiffies to increase and it is the software
watchdog that kicks the server over.

On the other hand with SMP disabled 2.0.31pre9 seems very solid on all
the servers I have tried.

Jon. 

-- 


Jon. <jon@gte.esi.us.es, http://www.esi.us.es/~jon>