REJECTED: (S) ret=0x081c97fc but we wanted ret=0x082041ec!

[Mithreas]

Arelith has recently started up a 4th server, with a few minor system changes in the NWN scripting - nothing that should impact NWNX. However, in the week since launching this server live (and never in testing with small numbers of people), we've twice had the server get into a state where no module events seem to fire. Conversations don't work (even ones where a PC talks to themself), spawns don't show up etc. Heartbeats still happen as usual (except that items created on PCs during heartbeats fail), and PCs can move between transitions, but connecting players appear to have empty vaults and DM clients can log in but then get a black screen when trying to load the first area.

The only hint I have as to what's going wrong is that when the problem starts happening, nwnx2.log starts to generate a lot of these lines (one per NWNX command? Unsure).
REJECTED: (S) ret=0x081c97fc but we wanted ret=0x082041ec!

We're running on Linux with nwnx_chat, nwnx_funcs, nwnx_tmi, nwnx_names, nwnx_fixes, nwnx_odbc, nwnx_ruby, nwnx_events and nwnx_resetplugin loaded.

I'm guessing something has got corrupted in the in-memory executable (memory scribble?) and this log is NWNX complaining that it can't hook into the address it expects. But I'm hoping that one of you with more expertise in this area can give me some leads I can follow.

Many thanks,
-Mith
_________________
Admin team, Arelith PW
Hakless world that relies heavily on the great work done by the NWNX team.

[Baaleos]

Im not familiar with linux nwnx.
But when win32 nwnx starts up, it usually prints a message to state whether the plugin started correctly - do you get these messages?

I would start removing plugins in order of risk.

Eg- Oldest plugins - keep,
newer ones - remove.

Remove them one by one, until the issue goes away.


It does sound like its trying to hook onto a memory location, but its ending up at the wrong place.


Strange thing is that I got the latest linux svn and I cant even find the word 'REJECTED' in the source via windows search tool#
Either windows is being a tool, or the word Rejected isnt in the source?


Is there anything in your modules onLoad event that might be firing in a particular mod?

[Mithreas]

Hm. I've done some more investigation and monitoring and this log is showing up on all of our servers, so it's not correlated with the module getting into this bad state.

Back to the drawing board... I'll worry about this log when I find out what function is broken and thus have a better lead!

Thanks,
-Mith
_________________
Admin team, Arelith PW
Hakless world that relies heavily on the great work done by the NWNX team.

[addicted2rpg]

Hello Mithreas, I am sure you remember me Its nice to see Arelith is coming along as it is.

I'm sure that your assumption about some corrupted memory is correct. I may even be so bold as to say that the bulk of your description is about spawning and that would at some point lead to the server trying to allocate more memory to its heap. I would imagine the OS call to malloc() or some such is succeeding, but it may be failing to try to integrate the new object into the game's internal record keeping system.

All that being said, it is kinda chilly that the log entry is showing for all of your servers but they continue to run anyway (except your 4th). For good practice, its obviously something you want to go away even if so far it seems benign. While I am sure I cannot get more RpM for this answer, I think your best bet may be cross-checking versions. For example, on the primary version you may be running version X of nwnx_fixes, while on your 4th server you may be running latest version Y of nwnx_fixes (I'm just picking on nwnx_fixes as a random example).

Another words, there might be new stuff up that would otherwise break your primary server, hehe..... So be cautious. Have you tried doing a direct copy of the modules for your 4th server?