Edwin Pratomo wrote:

>>which brings me to another question: is rlimit-ing ibserver safe? (i.e.:
>>would it corrupt databases in the case of abrupt process termination by
>>the OS?)
> I don't know, but IMO, it should be safe. A failure to allocate memory
> should be handled in a safe way, not to corrupt the database.

i'm sorry. i had thought that address space/data/stack size limit works
like cpu time limit. of course it should be safe.

> Also the
> behaviour may be different for SuperServer and Classic. This question is
> better answered by someone knowledgeable about the Fb internal.

i think rlimit will work in both cases. the only difference is that,
classic is forking a new process for every new connection.

> Frankly, I was thinking about SERVER_WORKING_SIZE_MAX configuration
> parameter when replying to your message. I've just remembered that the
> InterBase's Operation Guide says it's only for Windows NT environment.
> But, since it's been quite a while since the Op guide was written,
> perhaps that has changed, or may be changed in future. I don't see why
> such parameter is only applicable to Windows.

i guess that's because the configuration uses the
SetProcessWorkingSetSize API function, which is windows specific?

--
sh