Tweet
The server has now been up for nearly five days straight now, over 57 different characters logged in over this time.
In that time:
That is the part we care about most. The process is not falling over.
What we are still chasing is the last bit of tick lag — the occasional hitch where the main loop takes longer than the 20 ms target.
Most ticks are fine.
Every so often we still see a spike (one stall of about two seconds on this run, and a few shorter ones around half a second).
You might feel that as a brief freeze or a timeout if you were unlucky enough to be connected at that moment.
That is not a crash. It is a performance leftover, and it is the next thing on the list.
We have just added a full tick trace on the server: every loop is timed, slow ones are logged with the cause (GC pause, SQL, object processing, and so on), and we can scroll the last couple of thousand ticks instead of only seeing a “worst ever” number.
Next reboot will run with that in place so we can pin down whatever is left and kill it.
Keep posting anything that feels like a freeze, rubber-band, or sudden disconnect — times (UTC if you can) help a lot now that we have the logs to match against.
Thanks for stressing it. The bones are holding.
The new server screen below gives us a lot more details about that is going on
.
Next update in a few days when we have collected enough logs to find the remaining issues.
In that time:
- Zero crashes
- Zero unhandled exceptions
- The server stayed up — maps, monsters, logins, the lot
That is the part we care about most. The process is not falling over.
What we are still chasing is the last bit of tick lag — the occasional hitch where the main loop takes longer than the 20 ms target.
Most ticks are fine.
Every so often we still see a spike (one stall of about two seconds on this run, and a few shorter ones around half a second).
You might feel that as a brief freeze or a timeout if you were unlucky enough to be connected at that moment.
That is not a crash. It is a performance leftover, and it is the next thing on the list.
We have just added a full tick trace on the server: every loop is timed, slow ones are logged with the cause (GC pause, SQL, object processing, and so on), and we can scroll the last couple of thousand ticks instead of only seeing a “worst ever” number.
Next reboot will run with that in place so we can pin down whatever is left and kill it.
Keep posting anything that feels like a freeze, rubber-band, or sudden disconnect — times (UTC if you can) help a lot now that we have the logs to match against.
Thanks for stressing it. The bones are holding.
The new server screen below gives us a lot more details about that is going on
.Next update in a few days when we have collected enough logs to find the remaining issues.