[442] in testers
Print scheduling
daemon@ATHENA.MIT.EDU (daemon@ATHENA.MIT.EDU)
Sun Dec 10 00:52:54 1989
From: probe@ATHENA.MIT.EDU
Date: Sun, 10 Dec 89 00:52:17 -0500
To: bug-palladium@ATHENA.MIT.EDU, testers@ATHENA.MIT.EDU
Cc: ops@ATHENA.MIT.EDU
Reply-To: Richard Basch <probe@ATHENA.MIT.EDU>
I am concerned with the fact that the scheduling is done to the first
available printer. This is fine until more than one printer is free.
If the search is done sequentially, then the printers that are first in
the search will tend to be more heavily excercised and run out of toner,
or otherwise break-down under normal wear, sooner than the other
printers.
The searching for a printer should probably be done on a pseudo-random
search sequence. Perhaps, it could be made into something as simple as
a sequential search, whose search begins at the next printer in the
sequence every time a job is submitted.
I am just trying to present the operational concern of uneven wear and
degradation of overall print quality by tending to submit jobs to a
printer that essentially has a higher preference simply due to the order
in which the supervisors are started, which will no doubt be fairly
constant between reboots.
-Richard
