Re: more question about porting/gcc

[ebertbd@tick.informatik.uni-stuttgart.de (Bernd Ebert)]

Hi

>This just sounds wierd. MiNT gives few slices for newly started program 
>to Mshrink until other process' are executed. My friend has been running 
>bbs few years now under MiNT with three nodes (one local) and he is able 
>to do almost everything (including gcc compilation) while there's users 
>logged in. No problems with out of memory errors or program crashes. I 
>doubt any deamon is writen so badly that it crashes when there is no 
>more memory left. How do you set the caches on? There's been some 
>problems with the old versions of hades.acc.

I know of the problems with caches, Actually I'm the one who worked out the
solution together with the Developer of the Hades (Fredi Aschwanden).
The BBS you mentioned is not a problem with what I described. The problem 
occurs for e.g. when there is heavy cpu-load and program starting and stopping
(gcc with a big package, like samba) and then for e.g. another user trys
to log in via telnet. the inetd trys to start in.telnetd but this is not possible
due to lack of ram and then inetd sometimes crashes.
When you try to start a second compilation (for e.g. mint-kernel) then
you will see what I mean.
This only occurs when a second program trys to start before the last started
process runs its mshrink-command.

>
>> But after all, my hades had no serious system crash. Once up and running
>> it remained up until we shut it down in the evening, this means 2 Users
>> working about 14hours with heavy cpu-load and with many applications
>> running. :-)))) 
>
>Even a week is nothing if not quite MiNT aware is executed. Before my 
>friend bought a hades he was running the bbs with falcon. The best uptime 
>was something like two weeks with MiNT and N.AES. The reason for the boot 
>was not crash but some configuration change.

I know that Mint is running rock stable, I'm running a small bbs on my
own with my hades. But that is not what I meant. When we worked we had
approximately 70-90 Processes running the same time and there was a memory
fragmentation of up to 97 Blocks. The system still worked fine. 

Greetings,
				Bernd