attachment
<html><head><meta http-equiv="Content-Type" content="text/html charset=us-ascii"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space;" class="">Ira,<div class=""><br class=""><div><blockquote type="cite" class=""><div class="">On Nov 26, 2014, at 11:51 AM, Ira McDonald <<a href="mailto:blueroofmusic@gmail.com" class="">blueroofmusic@gmail.com</a>> wrote:</div><br class="Apple-interchange-newline"><div class="">Hi Mike,<br class=""><br class="">For operation (3) DeregisterOutputDevice, I was thinking of the case<br class="">where a running service (Print in the near term) crashes (and this is<br class="">noticed by System Service, let us hope). If DeregisterOutputDevice<br class="">is (redundantly) supported by the System Service, then an Administrator<br class="">could do the cleanup for the Infrastructure Printer (who might queue<br class="">jobs).<br class=""></div></blockquote><div><br class=""></div>Presumably DeregisterSystem is the way for the System service to clean up references to the output device in all services it manages, even those that have "crashed" (which is a state we don't really model semantically...) If you want to deregister from a single service, you talk to that one service.</div><div><br class=""></div><div><blockquote type="cite" class=""><div class="">Or, I'm crazy, and the System Service would automatically do this<br class="">cleanup internally and no external operation needs to be exposed.<br class="">But that risks a Deregister when the crashed service can be<br class="">successfully automatically restarted by the System Service after<br class="">a brief interval of unavailability. <br class=""></div></blockquote><div><br class=""></div>I would assume that self-healing of this sort would be expected of a cloud/SDN solution, with notifications sent to the relevant admin/operator. And I don't think we can necessarily recover from all situations like this - the Proxy might well have gotten a response from its Deregister-Output-Device request, with the cloud service crashing after successfully sending a response, so we have to hope that implementations will provide a way to perform manual cleanup as needed...</div><div><br class=""><blockquote type="cite" class=""><div class="">I have mixed feelings about the reliability (and advisability) of <br class="">automatic recovery by the System Service for crashed or stalled <br class="">Job Services.<br class=""></div></blockquote><div><br class=""></div>Typically servers will be restarted a certain number of times in order to provide continuous service, with an alert going out for each crash/failed restart. And such things are typically configurable by service and various events - I have such things setup with our <a href="http://PWG.org" class="">PWG.org</a> server instance, for example...</div><div><br class=""></div><div class="">
<div style="color: rgb(0, 0, 0); letter-spacing: normal; orphans: auto; text-align: start; text-indent: 0px; text-transform: none; white-space: normal; widows: auto; word-spacing: 0px; -webkit-text-stroke-width: 0px; word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space;" class="">_________________________________________________________<br class="">Michael Sweet, Senior Printing System Engineer, PWG Chair<br class=""></div>
</div>
<br class=""></div></body></html>