Hi there, to keep at the safe side, we bounce our application once a day. The app happens to be somewhat specific in the sense it is really used only locally and at the business hours; therefore we can afford simply to not-gracefully relaunch it late evening.
Worked well for a long time. Lately the load increased, app worked, but got slow. We simply added memory, and it helped — the heavier load is served quickly and reliably. What's weird is that the site administrators started to complaint that when there's a heavy load (served all right) during the day, the application does not relaunch properly at the evening. It looks like it hangs in some weird way; the admins have to notice and Force Quit, which reportedly works all right. One thing is the question why not-graceful schedule might fail; we are investigating, looks like perhaps sometimes (for reasons yet unknown) a worker thread might lock (in a weird way similar to an endless loop, for we found no deadlock), and that might prevent the relaunch. The main question why I write: isn't there a way to tell wotaskd or JavaMonitor or whatever to try the normal schedule, but if it fails to re-launch properly in, say, half an hour, to automatically force quit and re-launch then? (And, preferably, send an e-mail to admins.) I sort of recall I have read somewhere the Wonder wotaskd+JM should be able to do just that, but can't find the source whatever I try, nor I can find the appropriate option anywhere. (Truth is, deployment is not exactly my cup of tea and I can well be missing something of importance. Looks like, alas, the site admins still ask me, for I “know WebObjects” :)) Deploying on CentOS, if that is important. Thanks, OC
_______________________________________________ Do not post admin requests to the list. They will be ignored. Webobjects-dev mailing list (Webobjects-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/webobjects-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
