Re: PROPOSAL to serialize cron.{daily,weekly,monthly}

[Abdullah Ramazanoglu <ar018@yahoo.com>]

begin  Carlos Sousa <csousa@tvtel.pt> dedi ki:
> On Sun, 05 Sep 2004 21:23:38 +0200 Thomas Hood wrote:
>> On Sun, 05 Sep 2004 20:10:07 +0200, Andrew Suffield wrote:
>>> On Sun, Sep 05, 2004 at 06:54:34PM +0100, Carlos Sousa wrote:
>>>> On Sun, 5 Sep 2004 10:47:35 -0500 Steve Greenland wrote:
 --8<--

>>>> I would say this serialization is a good idea and worth the effort, on the
>>>> basis of system load and machine stress considerations alone.
>>> 
>>> The opposite is true on hardware capable of true parallel
>>> operation. So once again that appears to be a matter for local
>>> configuration.
> 
> OK. But then there is still the problem of conflicting concurrent jobs...
> 
>> There remains the question of what the default should be.
>> 
>> Would serialized execution be significantly better for the vast
>> majority of systems out there?  If so then it might be worth
>> implementing the idea.
> 
> All major consumer pc stores here carry mostly (only?) single-processor
> machines. All computers I personally know about (mine, friends', of schools
> and other institutions/companies I've worked in or visited) are also in
> this category, and most are even reasonably recent machines. We're still
> a bit far from parallell hardware for the masses ;)

Also, more often than not, maintenance scripts are more I/O bound than
CPU. So an SMP system wouldn't help much. Also I/O parallelism needed.
Such a parallel system is an exception to the rule, and it will certainly
have a skilled admin to take care of everything. IMHO a default system
should be geared for resiliency and fool-proof operation, even in the
hands of a non-competent admin.

-- 
Abdullah        | aramazan@ |
Ramazanoglu     | myrealbox |
________________| D.0.T cöm |__