zwicky@erg.sri.com (Elizabeth Zwicky) (04/02/91)
In article <2147@brchh104.bnr.ca> de@helios.ucsc.edu (De Clarke) writes: > But I didn't think you needed separate spool directories for >remote printers. Am I in left field? You are in left field. A printer job, as a file, does not contain any information about what printer it has been spooled to. When it is spooled, an lpd starts up, and that lpd gets told by lpr what the relevant printer is. As long as that lpd is running, it will spool the job to the correct printer, no matter how the spool directories are set up. (It will also drain all other jobs in what it thinks is its spool directory, but you are unlikely to notice this unless printing is heavy and directed to multiple printers.) When you kill off lpd and restart it, it reads through the printcap, and checks for outstanding jobs in the queues. The first printcap entry for a given spool directory will get all of that directory's entries. The reason that people often let all their remote printers use the same spool directory is that they assume that all jobs will drain rapidly. When this assumptions fails, so does the printer system. (By the way, the easy way to undo the hang when you get a "waiting for remotehost to come up" message is to use lpc to abort and restart the relevant queue.) Elizabeth D. Zwicky