Thralls are being sent undermesh by the system and deleted

[Toushou]

Basic Info:

Platform: Steam
Issue Type: Gameplay
Game Mode: Online Private
Server Type: PvE
Map: Exiled Lands
Server Name: Exiled Mimions
Mods: Emberlight, Northern Timber, AG - Architecture, Male New Faces, Female New Faces, GetClose, Agony’s Attic, Barbarian Barber, GrimProductions, Cookie’s Skinnery, Devious Desires, Cookie’s Immersive Armor Reupload, Juvilia’s Crabby Fasion, Juvilia’s Crabby Fasion Vol.2, Juvilia’s Crabby Accessories, Exiled Aesthetics, Emu’s Cat Variants, Fog Reducer, Martial Combat, (REFORGED) Yamatai Exiles - Assassin’s tools (Ninja weapons), (AGE OF WAR) Yamatai Exiles - Katana Family, Yamatai Exiles Patch, Evils Cabinet, JAS: Juvilia & Agony’s Salon, Extended Shortcut Bar, Sikk’s Clan Emblems, Night’s Calvalry Armor Set, Less Building Placement Restrictions, LBPR - Additional Features, Pythagoras: Expanded Building, Invisible Warpaints, Max Level 70 (Harder), Sorcery Tweaks, Tot ! Sudo, Tot ! Admin, Tot ! Chat, Tot ! Custom

Bug Description

Thralls are being sent under-mesh by the system and deleted

Bug Reproduction:

Hard to reproduce, but I suspect that the bug occurs when loading in the area the system snaps the thralls into the standing position while things are loading in causing a position error that sends them under the mesh. There’s a delay from the glitch occurring to the time the thrall is deleted, likely from falling.

I’ve got some samples from the event logs, taken on different days when I can catch it, and the server-log is from the 4th picture, the one with the thrall marked (Bee). I logged off right after it happened and pulled the log in case it helps.

ConanSandbox.log (1.3 MB)

[TackleBury]

Greetings, I’ve had something similar happen to me today. I grabbed two of my thralls from the point I went down, logged off at base, and now those dudes are gone. My follower count has decreased. I can’t find them on the list. My log had nothing in it. No sign of death or decay or whatever…. frustrating. What are your thoughts? Same bug?

[Toushou]

Normally, a thrall you have following you when you log out will return to its assigned “home“ be it a point they guard, or the last object you or the system assigned at “home“.

Keep an eye on your event log as the time of disappearance and deletion can differ. you may be able to at least locate where in your base the thrall was by looking for the last place it was before being deleted.

To do this start with having only Thrall/Pet checked days past set to one preferably, two at most since the event log doesn’t keep a long list, and work down the Proximity and check the living settlements box until you find where the thrall was trying to go before it was sent to the void. This can be harder if you have a lot of active thralls moving around and/or have duplicate names.

Sometimes moving things around to ensure a clear path to an object can help minimize losses from what I’ve experienced.

In my example pictures above, I’ve highlighted their last location in green, the location highlighted in red is roughly the same on the X/Y axis but way lower on the Z axis indicating that they fell through the world. Its interesting that it gives the location after being deleted.

[Toushou]

Update: While testing the server with Living Settlements turned off, I lost multiple thralls in short order, the system is still sending them into the void when seated while loading in.

An interesting side note, this Merram the Cook was one that was lost a while ago but somehow returned in the void downed. I couldn’t rescue them due to botching another rescue attempt, it doesn’t work while flying but will still start the 1 hour cooldown timer.

I also confirmed that the thralls are dying at the Z -1,048,000 range, as it also kills players.

Since the logs sometimes say a thrall is placed above the kill range but still undermesh, I think what’s happening is that when the game is loading an area with a seated thrall it attempts to attach it to the seat it was in when the area was unloaded, if the seat isn’t also loaded at the time of the check the game tries to place them on the ground, but if the ground isn’t loaded under the thrall at this time the game will continue to look down until it hits the limit and throws the thrall into the void.

Perhaps changing the load in priorities to have entities like mobs, thralls, player, be loaded after world objects and player placed objects could fix this.

[Toushou]

The server IS modded. I’ll do further tests on a un-moded single player map tomorrow, its late for me at the time I’m posting this.

[Toushou]

Update: single player bug testing

Setup: I downloaded the server’s Game.db file and started with mods remaining on as a control, same test environment, just in single player. With with living settlements turned off I spawned in and seated dozens of cook thralls to maximize the odds of the bug occurring.

Test 1: Spent a few hours loading and unloading the area, no losses.

Test 2: Maxed out graphics setting to induce lag, no losses.

Test 3: Living settlements on, laptop graphics to remove lag variable, no losses.

Test 4: Living settlements on, max graphics, no losses.

Test 5: Disabling living settlements again and re-seating thralls to test if there’s any lingering affect from living settlements being on. 2 losses, a DNBO rather than an immediate deletion. unable to rescue due to them taking damage, dead before the cooldown for damage taken was cleared. no further losses during testing period.

Results: The DBNO losses are not conclusive to being the same bug that sends thralls into the void, though both send thralls undermesh the resulting deaths are different.

Perhaps communication delays between client and online servers is a factor?

Needs further testing to replicate the bug before moving on to removing mods.

[Toushou]

Thank you for the info, the bug is on the multiplayer server I run and I pulled the game.db file off of the host site to test if the bug could be replicated in single player. I’d like to be through so as to give the Funcom devs as much info as I am able to provide in narrowing down causes.

I’m in the middle of retesting single player at this time, giving each test more time, before I’ll test the multiplayer server again.

[Toushou]

Update: single player bug testing (day 2)

Setup: starting from a saved Game.db copy of the previous setup prior to testing.

Between each test I’ll also be reloading the same saved game.db file to remove previous test variables excluding test 5 which will immediately follow test 4.

All five tests were run for four hours with no losses.

online server tests in progress.

[Toushou]

After lengthy tests I’ve confirmed the bug does occur without mods and only in online play with Living Settlements on. Replicating the bug occurs when loading or unloading an area with a high volume of player placed assets (building pieces, decorations, crafting stations, ect.) thralls being loaded/unloaded may not be properly set at ground level and will fall undermesh and die, this may take several minutes to register in the event log due to the thrall falling to the void.

I’ve also determined at least three variations of the bug.

1. Thralls are killed by Unknown (the void) with no other info given. This variant is the one that I confirmed without mods.

2. Thralls are killed by Unknown (the void) with a following message of being moved to the Z -1,048,000 range by the system. I believe this is caused by the Pythagoras Expanded Building mod.

3. Thralls fall and are DBNO and unrecoverable from taking fall damage, eventually dying. This is the least common variant and I have yet to pinpoint if a mod is the cause, but in extremely rare cases (one time in my tests affecting multiple thralls) this can affect fighter and archer thralls, so this may not be related specifically to the Living Settlements system.

This concludes my testing, is there anything specific I need to provide to help Funcom verify this for their dev team?