Briggs Softworks

10 min

escorpius wrote: ↑Thu Sep 01, 2022 5:17 pm10 min

Change it to 2 and see if that makes any difference.

We will.

We just experienced an extended outage as our admins were unavailable for a period of about 20-25 minutes. It's safe to say that [a] the server never recovers and the PM_timer thread dies without any stack trace, error logging, etc.

The latter seems very odd to me, particularly given what you said about expecting there to be a stack trace in the error log. What could possibly be causing a quiet death on that thread? Is there any way to ensure that this thread dying actually logs something so we can begin to unravel this issue???

escorpius wrote: ↑Thu Sep 01, 2022 10:48 pm What could possibly be causing a quiet death on that thread?

Don't know (yet) but I'll be looking closer at that clean-up code especially if changing it back to the default setting makes a difference for you.

Thanks. We'll keep an eye on it and let you know if we run clean for a while.

The change on the timeout has not stopped the behavior. We have built a watchdog that now monitors for that thread to die and then alert, which we'll soon have automatically restart the service. This is far from ideal, but at least minimizes our downtime, in theory -- although I believe your server still requires a manual restart through the admin, i.e., the API doesn't support stop/start, yet.

That said, I would still like to understand how that thread is dying without an error being logged. You mentioned moving to a new logging approach -- might it simply be failing? Or, is there something unexpected in your code path? Having a stack trace would be extremely useful. Have you made any progress on understanding this?

escorpius wrote: ↑Thu Sep 08, 2022 7:49 am although I believe your server still requires a manual restart through the admin, i.e., the API doesn't support stop/start, yet.

There is an "Auto start at bootup" setting but it requires an actual reboot, not just an manual service start.

That said, I would still like to understand how that thread is dying without an error being logged.

That makes two of us.

escorpius wrote: ↑Thu Sep 08, 2022 7:49 am That said, I would still like to understand how that thread is dying without an error being logged.

Please contact me via email with your license key and mention this thread. I've prepared a beta update with extra error trapping in the timer thread and would like to see if it gets triggered.

Kent Briggs wrote: ↑Thu Sep 08, 2022 9:42 am There is an "Auto start at bootup" setting but it requires an actual reboot, not just an manual service start.

Correction: manually starting the service by any method other than the PM Service Manager will cause the poker server to go online if that setting is enabled. From a command line/batch file (under Windows Administrator privileges):

net start PokerMavens7

The service must have already been registered/installed. That can also be done from the command line:

PMService.exe /INSTALL /SILENT

Ah. That would explain some of the differences I'm seeing between our automatic restart of the service, which is done through Task Scheduler, and when I restart manually using the Service Manager because of a crash.

Thanks! We'll test this out.