For the DEVS - 'You must wait to use something again'

[OldnGrey]

With this new core I am suddenly getting random reports from players getting the pol.exe core message "You must wait to use something again".

I don't see anything in the core-changes remotely likely to be the cause. Trouble is, it might be a core breaker for me unless I can get to the bottom of it.


UPDATE.
I have rolled back to RC5. This isn't good at all. The core message took hours to stop for the player, and was uneffected by logoff. Also once it happened ALL players stopped getting regen of any sort.

I hope it's me, but I will report back when I know more. It was vital to give the players a working shard again.

[CWO]

Do you possibly have

AssertionFailureAction=continue

in pol.cfg? If so, this could cause exactly the problem you're having. If you don't have it at all, then its abort which is fine unless you want to save on assertion failures before the shard crashes.

[OldnGrey]

Yes it was set to continue.

I will do some testing with it set to shutdown, but a check of the logs showed there were no errors recorded for assertions or exceptions.

[OldnGrey]

I am now 100% sure that my scripts are the same between the 097 2008-02-26 core and 097rc5. A directory tree comparison using 'Ultra Compare' shows identical cfg src and inc files before compiling each set.

I have had no problems since reverting to 097rc5.

I am still unable to reproduce the problem on a test server, but I will keep trying. It's either triggered by the number of players or by an event that nobody has been able to report accurately. There were no logged errors at all in pol.log, scripts.log or debug.log.

The message "You must wait to use something again" is a pol core message. We were only live with this core for around 12 hours and it didn't have this error initially. But more and more players started getting the error message and they were unable to continue playing despite logging off and on again. I asked everyone what they were doing before it went strange, but didn't get anything back that reproduced the problem.

CWO says that the problem with the regen hook not firing is the same as if there had been an assertion errror in the core although none was reported in the logs. Therefore I don't think the regen hook firing is a bug, more of a result of an assertion problem maybe.

Does anybody else using 097 2008-02-26 on a live shard have any problems like this? I'd like to get this sorted out but obviously I can't fix a core problem.

[OldnGrey]

If nobody else is having the same problem I will have to assume that others are using the core but it's not causing the problem.

After a much more careful check of logs I think I have overlooked a situation that appeared just after we cutover to the new core.

The USA had daylight saving change and the system clock jumped forward one hour.

I wasn't watching the console, but in the logs the cpu went to 100% for a while and the normal 1 minute log entries suddenly went beserk trying to catch up the hour in the entries.

Try it on a test shard. The console shows all manner of crazy errors that don't appear in the logs - all manner of thread and checkpoint stuff and no clock movement warnings.

I tried this on rc5 as well as 097 2008-2-26

I will try the core out on a live shard soon.

Anyone suspect this might be the cause of the problems before I try it out on the players? And if so have you ever had a problem when going across to daylight savings?

[CWO]

My shard didn't have a problem crossing DST although my shard is 096.4b

[OldnGrey]

I plucked up the nerve to cutover to POL 097 2008-02-26 again.
The timing could have been better because a major quest was about to start unbeknownst to me.

It's been around 20 hours since we've been on the new core and I am gradually de-stressing. There are no instances of the ''You must wait to use something again' messages again.

I can only conclude that the recent pol097 cores really do hate the system clock being advanced by one hour. I've tried it many times on a test shard, and seen all sorts of responses, no two the same - it depends on what scripts are running it seems. Going back 1 hour has no discernable effect.

Too early to tell, but so far I don't have the urge to cut back to rc5.

[*Edwards]

found in servspecopt.cfg:

#
# DoubleClickWait - Number of seconds that must pass before a player can
# double click an object again. (second)
#
DoubleClickWait=1

"You must wait before using something else again" is sent to the players if an item is used too fast.

[OldnGrey]

I had and have
DoubleClickWait=0

I put it down to a frozen thread in the core. It was a symptom, not the cause I think.