[Linux-ha-dev] Heartbeat 1.2.2-3: TTY Overrun on SLES 8 / 2.4.21-138

[Lars Marowsky-Bree]

On 2004-12-08T09:55:50, Alan Robertson <alanr at unix.sh> wrote:

> >At this point, we think that we need to reduce the FIFO interrupt 
> >trigger for the UART (currently set to 14 ), thus allowing the interrupt 
> >plenty of time to occur and hopefully getting rid of the overrun ...
> 
> That sounds like a good idea to me - provided you know how to do it ;-).
> If the FIFO is 16 chars deep (probable), then this seems a bit too much of 
> a near thing to me. But, even so, CTS and RTS should prevent this problem 
> completely with good cabling.  I'd set it to something more like 8 and see 
> if that helps.

I just want to reiterate: This is the first ever case of this being
reported. 

I'd still bet a $100 (USD, which is quite cheap for me to do from the
Euro zone ;-) that it's the cable or something in the hardware itself.
(As repeatedly stated within the associated bugzilla entry.)

Fudging around with the kernel will at best hide the problem to make it
reappear in the most inconvenient timing situation, causing the maximum
loss of money these systems are capable of. (Murphy's Law applied.)

But, I can also radiate a very strong SEP field, in particular now that
I'm supposedly on vacation. But, just so that I can later do "neener
neener told you before" ;-)


Sincerely,
    Lars Marowsky-Brée <lmb at suse.de>

-- 
High Availability & Clustering
SUSE Labs, Research and Development
SUSE LINUX Products GmbH - A Novell Business