CLUSTERQUE FAQ'S

 
Q If there are 3 printers in a cluster, large number of small print jobs should get evenly distributed over the 3 printers?
A Jobs will be sent to the available printers on a first available, least-used basis.

Q If one of the printers is OFFLINE, the job should not be sent to it at all?
A It depends on what the real network status of the printer is. If the printer is truly not accepting network connections (turned off, buffer full, etc) we will not send jobs to the printer even if the status of the printer on our queue says "Idle". However, the printer can look like it is off line, but still be accepting data -- this is an HP peculiarity, but we have no choice but send data to the printer if it is accepting data (for example, on my HP2100, if I open both trays on the printer, the 'OFFLINE' light goes on, but the printer still accepts data -- I suppose this is on the assumption that the next job may be requesting data from the manual input tray, so the firmware says it will not stop accepting jobs). The way our software works is to check two things before every job:
  1. We check the TCP/IP socket (port 9100) to see if it will accept data (this checks whether the printer is turned on.
  2. Then we send 1000 null characters to see if the printer is actually receiving data. If it is, we send the job
.  If you have selected the 'Query printer' option, then ClusterQue checks the printer as the job is sent to see if it actually printing the pages. If it is NOT, then at a certain point (hard to predict since it depends on the complexity of the print job, etc), the job status will change to either AsstReq (if the job is only partially sent to the printer) or NetError (if the job is completely sent, but no pages have printed) and the job will be put back on the queue and sent to another printer. The status of the printer will stay in the error condition until the problem is resolved (we check this by querying the printer for a code that says paper is loaded to every tray, etc) -- note that this is a more stringent condition than that required for starting the print job. 

Q The refresh button should immediately update the status of all printers in the queue?
A We do not check the status of any idle printer. We do however recheck the status of every 'Error" printer at 30 second intervals.
 
Q Suppose there’s only one job sent. It should go to the 1st printer in the cluster. Even if the 1st printer is now free after printing, the 2nd job should go to the 2nd printer so that there is even wear and tear on the printers in a cluster?
A Yes -- this is the way it works.
 
Q Please define the difference between various options:
  • Requeue
  • Delete after print, etc..
A There are four of what we call "dispositions": 
  • Delete -- delete the print file after completely printing
  • Requeue - leave the file on the print queue after printing. The file will not be removed from the queue or deleted without a manual action.
  • Remove - [the default behavior] remove the file from the queue, but do not delete the print file.
  • Leave - leave the job on the print queue for 5 minutes after printing, then remove the job from the queue (and delete it if requested). Note that the leave time is configurable in the System properties.

Q Also please explain the meaning of CLASS and how it can be used.
A Class is used as one parameter used to determine the physical printer that a job may print to. The classes do two things: if there is only one class on the printer, then only jobs with that class can print on the printer.  If there is more than one class on the printer, then jobs will print on that printer if the job's class matches one of the printer classes. The order of the classes assigned to the printer determine the priority of a given job: if the classes assigned to the printer are (for example) XDE, then class X jobs will print first, then class D jobs, then class E jobs. This is the way we preferentially print the larger parts of a split job on the faster printer.

The details of this are described in the Tutorial document.

 
Q Also how can a stalled print job be rerouted manually or automatically.
A To reroute a stalled print job manually:
  1. Go to the manager program and highlight the print job.
  2. Press the "Stop/start" button (status will change to "Stopped").
  3. Highlight the print job and press the 'Hold/release' button.
  4. Highlight the print job and press the 'Modify' button, then choose the new printer.
  5. Highlight the print job and press the 'Hold/release' button.
To reroute the stalled job automatically make sure the 'Respool jobs requiring manual assistance' options selected in the System properties specify the time in minutes). Note that there are two different cases in which this can happen:
  • The job is being sent to a cluster printer. Then the print job will automatically be put back on the print queue and will be sent to the next available printer.
  • The job is being sent to a specific physical printer. Then the job will automatically be sent to the backup printer for the stalled printer, if one has been configured. If there is no backup printer, the job will be put back on the queue, but there will not be anywhere it can print.

 

Q When I send many small print jobs (1-3 pages each) to a printer with ClusterQue, I see a 2-5 second delay between jobs.  How can I eliminate this delay?
A You need to uncheck the "Query Pages Printed" option from the printer configuration utility.  This capability is used primarily for page-level automatic recovery on large print jobs.