Re: taper algo change
2003-08-11 12:10:36
Largestfit() isn't some balancing act between the estimate size
and the capacity of the spool/holding area ?
What is the algorithm for starting the dumps anyway ?
ie: what DLEs does it attempt to dump first, is that based
on the taper algorithm ?
> Would it be possible for the driver not to queue up a tape job in
> startaflush()? I'm using LARGESTFIT, but for the most part dumps end up
> getting flushed to tape as they come in because there is often only one
> dump in the queue when startaflush() is called. I would like it if (at
> least with LARGESTFIT) that if there are more dumps to come in and the
> current dumps DON'T fit that it waits until the next dump finishes to
> try again. This would allow smaller dumps that come in later (due to
> various factors) to get onto the tape and make more effecient use of the
> tape.
>
> My proposed change would be something like:
>
> case ALGO_LARGESTFIT:
> fit = tapeq.head;
> while (fit != NULL) {
> if(sched(fit)->act_size <= tape_left &&
> (!dp || sched(fit)->act_size > sched(dp)->act_size) &&
> strcmp(sched(fit)->datestamp, datestamp) <= 0) {
> dp = fit;
> }
> fit = fit->next;
> }
> if(dp) remove_disk(&tapeq, dp);
> ! else if(dumps_left_to_do()) return;
> break;
>
>
> but I don't know what "dumps_left_do_do()" needs to be.
>
> --
> Orion Poplawski
> System Administrator 303-415-9701 x222
> Colorado Research Associates/NWRA FAX: 303-415-9702
> 3380 Mitchell Lane, Boulder CO 80301 http://www.co-ra.com
>
>
|
|
|