1.2.1: Repeated server crashes due to undefined "exception ticking world"

jpet

New Member
Jul 29, 2019
13
0
0
Version:
1.2.1

What is the bug:
I've been playing Direwolf20 1.7 modpack since 1.0.1 (1.2.1 now). During the first two weeks, the world was empty and all was fine. Then, as the map grows, i started experiencing those kind of crashes serverside
Sometimes, it happens only every other day but it can also crashes 2 minutes after a reboot.
DW20 1.7 version : from 1.0.1 to 1.2.1
JAVA : information can be found in the crash log.
Server Ubuntu : 12.04.5 LTS 64 bits​
Details :
  • I monitored the JVM and it doesn't seem that there's any problem with memory or cpu (except that computercraft is using a lot of cpu time due to sleep() command but i think it's a normal behaviour with CC). I don't see any performance issue serverside or clientside (25-60 fps).
  • When i see a crash, i can usually find a solution using the logs, but this crash log doesn't provide any high level data. We can only see that the higher level function involved in the crash is "net.minecraft.server.management.PlayerManager.func_72693_b" which is a base function that can be used by any mod (from what i know). So it doesn't help a lot... I see that a lot of people had experienced crashes on this function, but in that case, the log always tell them which mod is calling this base function.
  • As i had a crash (due to disk space) a while ago, i think it could have generated chunk inconsistencies on the map so i used MCEdit to "repair" the map but it hasn't helped.
  • All my machines are working fine except AE sometimes (strange energy consumption or import bus not working) but it seems to be mostly related to item routing and it's solved by now.

Mod & Version:


Paste.feed-the-beast.com log:
http://pastebin.com/4aPfpGQX

Can it be repeated:
I cannot find a way to reproduce : it can happen when i fly, when i search an inventory, when i'm akf (in any dimension). It seems that the server crashes only when someone is connected (not only me, it seems to be any player). A simple restart of the ftb server is needed after the crash, the world isn't damaged.

Known Fix:
A simple restart of the ftb server is needed after the crash, the world isn't damaged.
 

Trisscar

New Member
Jul 29, 2019
974
0
0
Huh. You know, someone else just posted about how one of the mods is showing as "Unknown by Unknown" for them. Wonder if this is related somehow, in terms of there not being much of an address for the error....
 

jpet

New Member
Jul 29, 2019
13
0
0
i checked it but the post you speaking of seems to be client side related.
(And i don't see any unknown mod in my config)
 

Trisscar

New Member
Jul 29, 2019
974
0
0
Eh, worth a try I suppose.
Sadly, I don't know much about servers, most of what I DO know about the console logs for FTB has been gained through logic and looking things up, not to mention the nice clarity that whoever coded the console has given. But despite that as you have said, this log is not very helpful.
...I think a way to find out more about this is to learn what that function IS that's being called, the one marked "func_72835_b". However, that would likely require someone who has actively coded a pack or mod, and there I cannot offer much assistance.
A quick search seems to link the function to generating structures or terrain that become bugged, but those are very different crash reports that aren't of this pack.
Long story short, MCEdit/delete the chunks that aren't important to you, then carefully explore till you hit something that crashes you again. Then view that chunk in MCEdit or similar and see what generated I guess. Hope it goes well. XD
 

jpet

New Member
Jul 29, 2019
13
0
0
Thank you for your interest but i don't really want to delete chunks randomly and redo all my world.
This issue cannot be resolve at my level. As i cannot locate the pb (crashes happen everywhere for my player and they are server-side only), I would have to shutdown everything and reboot my sub-systems one by one, including the wait before each crash (can take hours for 1 test and there's too many things to check). The problem here is that a mod desn't catch an exception correctly, i think. I've already spent too much time on this bug and i'm only hoping for an update that would adress this issue. If i get a readable log soon, i will be able to remove or redo the sub system causing it. If i don't get an answer, i think it will be easier for me to start a new world.
 

jpet

New Member
Jul 29, 2019
13
0
0
2 days ago, i realized that i had lost my +10 red heart bonus. I had previously removed the canisters from TiC player inventory but i thought the bonus was permanent. So i'm not sure if something weird happened there.
The point is that, without my +10 heart bonus, the server didn't crash for 2 days(it hasn't happened for months)! I re-added 1 heart canister -> a few minutes later, the server crashed.

I'm now looking for informations concerning TiC bugs or a possible conflict with another mod.
 

jpet

New Member
Jul 29, 2019
13
0
0
I removed the health modifier corresponding to my heart canister with an nbt editor and restart the server. I don't know if it's technically similar to what happened to my player (lost all my red heart canisters bonuses) but i didn't find a command for that and i needed to confirm that those heart canister were the origin of my pb.
Bad news : the crashes persist (~1/h) :(

Another thing i noticed, is that when i first connected appdynamics to my server JVM (monitoring application i use to help me diagnose the issue), all my turtles/computer failed to boot. I imagine it's due to a calibration process appdynamics launched against the jvm and that could have caused a lot of lag at the jvm startup. Anyway, all the turtles were OFF and i only needed to open the gui for them to reboot correctly. But it took me some time to notice that it affected all the turtles, so the last turtles were only reboot recently -> not long before the crashes came back ...
So i will keep looking around the turtles/computers. I'll let you know...
 

Trisscar

New Member
Jul 29, 2019
974
0
0
Huh, I wonder why the game wouldn't show the turtles and such calling things as they run... Odd. Course it might not be them at all, but still.
 

jpet

New Member
Jul 29, 2019
13
0
0
OK. I think i found it \o/
I've disabled a single turtle and i had no server crash since.

That turtle was condensing items in a chest every 8 seconds. The chest is managed by steve factory manager (ingot <> blocs storage & conversion). There's also a filtered item conduit (bi-directionnal) connected to this chest.
It looks like a concurrent access error. The origin would be one of those three blocs : the computer, the machine inventory manager or the item conduit.

I won't be able to do much more but if someone want to fix this issue, i can provide more data.