I ran into this today. I have a watchdog on the Pi watching for various conditions and hadn't been looking for it previously. Stumbled onto after about 4 hours stuck in this mode - by looking at the WU site.
I'm going to suggest that
at least part of the issue may be within CMX. That is based on just one experience which is hardly a pattern. After reading this thread, and monitoring a growing number of mono threads (336) and CPU usage, I tried a reboot. As far as I can see, after the reboot the WU updates are working normally, where prior to it they seemed to be failing with "WU update: The Task was canceled" consistently.
This is a fragment of the 2 sequential logs grepping for "
WU"...
Code: Select all
20180108-210151.txt:2018-01-22 09:47:40.619 WU update: The Task was canceled
20180108-210151.txt:2018-01-22 09:49:40.648 WU update: The Task was canceled
20180108-210151.txt:2018-01-22 09:51:40.702 WU update: The Task was canceled
20180122-095424.txt:2018-01-22 09:56:01.178 WU Response: OK: success
20180122-095424.txt:2018-01-22 09:57:00.744 WU Response: OK: success
20180122-095424.txt:2018-01-22 09:58:00.744 WU Response: OK: success
I also note a different pattern here when CMX appears to drop into the "WU update: The Task was canceled" mode. CMX does an excellent job of starting updates at the
top of the minute - and my WU updates fire once per minute. However, once this mode started, the "
WU" lines are 2 minutes apart. If I had to guess, I suspect there's a timeout at around 100 seconds, or 1:40 when the canceled message is issued. CMX likely starts the next attempt at the
top of the minute (20 sec later) and then takes another 1:40 to fail (skipping one attempt in the process). In the log the stride between errors is 2 minutes, but between successes is 1 minute.
Looking back through the logs, grepping for "
WU update" I note that there hadn't been a burp in about a week...
Code: Select all
2018-01-15 16:05:00.942 WU update: at System.Net.WebConnection.HandleError(WebExceptionStatus st, System.Exception e, System.String where)
2018-01-15 16:08:00.919 WU update: at System.Net.WebConnection.HandleError(WebExceptionStatus st, System.Exception e, System.String where)
2018-01-22 05:57:40.987 WU update: The Task was canceled
2018-01-22 05:59:40.996 WU update: The Task was canceled
2018-01-22 06:01:41.021 WU update: The Task was canceled
2018-01-22 06:03:41.006 WU update: The Task was canceled
By the way ... Seems like, at least for my station, WU did something around 10 Jan that dramatically reduced the WU exception messages.
My Python watchdog already monitors Diags, so I'll add code to handle this case. Based on this single experience, it appears an unbroken string of "WU update: The Task was canceled" messages over 4 hours was cured by the Pi reboot (40 minutes and updates without a burp). If I'm ambitious enough, I may allow for some number of these in a row before I indicate that a reboot is needed. From what I read here, it seems there may be cases where it fixes itself.
Cheers,
Bob