Zone crashes due to Handle_OP_SenseTraps
 

I'm seeing semi-regular crashes of zone.exe with nearly identical logs. In the eqemu_debug_zone for the same zone process, I see regular entries near the end that show NPCs attacking each other. I've never seen a crash that didn't include these same NPCs in it. There are usually other log entries after it, but not many.

Source Code is from 10-11-2014. Windows platform.

Pastebin of the log: http://pastebin.com/CFgy792G

[11.04. - 23:49:12] AIYellForHelp("Firiona_drixie016","goblin_raider00 0") Firiona_drixie024 attacking goblin_raider000 Dist 7082.875000 Z 205.830002
[11.04. - 23:49:12] Succeeded in setting inventory version to EQClientUnderfoot(5)
[11.04. - 23:49:12] Member of group 12002 named 'Minriu' had an out of date pointer!!
[11.04. - 23:49:14] AIYellForHelp("Firiona_drixie003","a_drolvarg_grow ler049") Firiona_drixie011 attacking a_drolvarg_growler049 Dist 9843.062500 Z 44.450001
[11.04. - 23:49:14] AIYellForHelp("Firiona_drixie024","goblin_raider00 0") Firiona_drixie038 attacking goblin_raider000 Dist 8589.366211 Z 196.080002
[11.04. - 23:49:21] Succeeded in setting inventory version to EQClientUnderfoot(5)
[11.04. - 23:49:21] Member of group 12002 named 'Rolt' had an out of date pointer!!
[11.04. - 23:50:10] The next weather check for zone: firiona will be in 600 seconds.
[11.04. - 23:50:11] AIYellForHelp("a_drolvarg_growler038","Firiona_dri xie003") a_drolvarg_growler003 attacking Firiona_drixie003 Dist 2719.414307 Z 39.080002
[11.04. - 23:50:11] AIYellForHelp("a_drolvarg_growler003","Firiona_dri xie003") a_drolvarg_growler006 attacking Firiona_drixie003 Dist 6101.682129 Z 15.447937
[11.04. - 23:50:19] AIYellForHelp("a_drolvarg_growler065","a_pilgrim02 3") a_drolvarg_growler032 attacking a_pilgrim023 Dist 5757.117188 Z 369.330017
[11.04. - 23:50:37] AIYellForHelp("Firiona_drixie008","froglok_ton_war rior000") Firiona_drixie004 attacking froglok_ton_warrior000 Dist 301.319366 Z 62.595268
[11.04. - 23:50:37] AIYellForHelp("Firiona_drixie008","froglok_ton_war rior000") Firiona_drixie040 attacking froglok_ton_warrior000 Dist 428.708771 Z 63.363953
[11.04. - 23:50:38] AIYellForHelp("a_pilgrim042","goblin_adept000") Firiona_drixie010 attacking goblin_adept000 Dist 9531.247070 Z 33.749969
[11.04. - 23:50:38] AIYellForHelp("Firiona_drixie010","goblin_adept000 ") Firiona_drixie018 attacking goblin_adept000 Dist 2875.682861 Z 9.095114
[11.04. - 23:50:46] AIYellForHelp("a_pilgrim003","froglok_ton_warrior0 01") a_pilgrim018 attacking froglok_ton_warrior001 Dist 121.528702 Z 147.457245
[11.04. - 23:50:52] AIYellForHelp("a_drolvarg_snarler014","a_pilgrim02 2") a_drolvarg_scavenger015 attacking a_pilgrim022 Dist 3232.700928 Z 70.489883
[11.04. - 23:50:53] AIYellForHelp("a_drolvarg_scavenger015","a_pilgrim 022") a_drolvarg_savage001 attacking a_pilgrim022 Dist 4968.435059 Z 64.887711
[11.04. - 23:51:02] AIYellForHelp("a_pilgrim003","a_drolvarg_savage004 ") Firiona_drixie028 attacking a_drolvarg_savage004 Dist 8903.751953 Z 204.174118

Well your line numbers dont match up and the crash dump was unable to get the lines so I can't tell what your issue is. So I guess update :P

Yes, I would say update, and ensure that you also copy the .pdb files that are generated into the folder where your zone.exe etc are. This should enable full logging.

Or if you don't want to update, ensure you have the correct .pdb files in that folder ( the ones that were generated when you compiled).

I just pushed out the lastest code as of this morning. If it happens again I'll put the logs up here. The .PDBs are there with this deployment.

There is a known issue with post-zone shutdown combat..and it's likely a result of that.

You will probably see lot's of crashes related to hate re-assignment.


Do you run on a vm per chance?

If not, do you have a lot of pagefile action going on?

In fact I am running the server inside a VM (hyper-v, not vmware).

That would make sense based on the zone debug logs. Seems entirely centered on Firiona Vie. It never seems to crash from any other zone. I'm considering adjusting the faction of the NPCs in the log to see if that makes a difference. I'm OK with some slightly inconsistent factions if it means not having a zone crash.

I'm sure it would crash anywhere there are mobs fighting..home cities, tutorial, etc...

Ok..the 3 cases I've heard of that have this problem all run vm's...


Any vm guru's have any idea why this bug would only affect those servers?


EDIT: Thanks for that, btw! :)

I doubt it has anything to do with being a VM, it sounds like a simple race condition and the VM may be more likely to trigger it due to slightly different timing.

Yeah, it is definitely related to agro and not specific to Handle_OP_SenseTraps. I saw another one last night and the offending method that time was Handle_OP_ShopPlayerSell. When I checked the debug log, it was when I was in Thurgadin and Coldain Warriors and Snow Cougars were fighting. Maybe a dozen lines later, the process crashed.

Here's another list of crash references that you may see as well: http://wiki.eqemulator.org/i?M=Pastebin&Paste=UfgWv1a5


Any chance these are script event calls being processed on mob entities post-consumption?