How SM crash system works?

Dragokas · **Author**

I noticed that under some circumstances, SourceMod exceptions that can be considered as uncritical can cause server crash.

I do like to know what these circumstances are, like:
- a certain number of errors per second (frequency) or so...
- whether this depends on the server settings, RCON start arguments?
- is this behaviour different for SM v1.7.3 / 1.8 / 1.9 ?

Like example: last time I wrote plugin with error "Out of array bound".
If this error occurred once, server is okay. But my cycle repeated several times per second and cause crash. Therefore, I came to the conclusion (possibly, incorrect) that the crash can be affected by the some frequency of exceptions.

Please, clarify. Thanks.

Kind regards,
Alex.

Fyren · 03-20-2018 , 18:57 Re: How SM crash system works?

The question is way too general. You're not going to get any kind of useful answer. If you want to provide reproducible ways to cause crashes, someone can look into why.

cravenge · *Last edited by cravenge; 03-21-2018 at 06:26.*

I kinda noticed this, too. Once SourceMod's error displays start to spam and it reaches a certain limit, the server would crash.

Dragokas · *Last edited by Dragokas; 03-21-2018 at 13:13.*

cravenge, exactly.

Quote:

- is this behaviour different for SM v1.7.3 / 1.8 / 1.9 ?

After we moved 1.7.3 -> 1.9, we noticed more crashes. 1-st day everything okay. 2-nd day - frequent crashes.
I can't reproduce now because we are rolled back.

There are can be many reasons, I know. I need read logs e.t.c. Analyze every plugin one by one ... Eliminate problems from hoster side.
But, Fyren, you are right, in this topic I want to talk in general, if someone of SM developer know:

Is lmit for crashes, cravenge was talking about, different in SM 1.7.3 / 1.8 / 1.9 ?

I know servers where they have thousands of errors, however, minimal freezes/crashes,
we have just few exceptions, but hard lags. Of course, it's plugin specific, so I'll continue to fix them.

Fyren · 03-21-2018 , 15:42 Re: How SM crash system works?

There's nothing in SM that cares about the number or rate of runtime errors.

I can come up with theoretical situations where aborting a script at a random point could crash or leak memory (without it being a bug in SM; a script could do something on the server that allocates outside of SM's handle system). A leak would only be a problem after enough memory is lost.