#Screeps Sandbox
5380 messages · Page 6 of 6 (latest)
there's a lot of unused code in there, things I was working on, or stopped using but didn't delete. a lot of tinkering.
and… running
I'm getting those from time to time: ```
Error: Isolate was disposed during execution due to memory limit
ooooh
memory limit is a different problem
something could be allocating infinitely instead of stack overflowing?
yeah, that's straight from IVM
though not sure if it's related to stack size or not
Actually scratch that, server was stuck in always reset mode
tick speed 1ms
it's zoomin'
1ms?? Lol
Yeah, it's just one sector
Maybe it's the docker server being much better than steam local server, I couldnt get ticks under like 500ms with the local server one
It's actually limited by how fast the client socket can pump out events
I dont know whats going on with the steam client but it's gone from working fine to just being unusably laggy even through restarts of the client and pc, I have it open now to push code and otherwise just use the browser
Damn, the moment it reached RCL3 it started crashing 😭
@arctic pendant It's complaining about structureRCLPositions being undefined
Also, if you wanna join on the pserver fun…
yeah, the room planner never ran for you
I've been manually enabling and running it
sorry that the codebase isn't in a grab-and-go state 🙁
Grab a copy of Jomik's server, put https://github.com/laverdet/screepsmod-inspector into the mod list and add 7777:7777/tcp to the list of visible ports
Then run the server as you would
localhost:7777 is gonna be a webpage with all the isolates and a link to the DevTools remote inspection link to it
I had to nudge Chrome a bit by opening a node --inspect first, then trying to connect to that
Then the IVM link worked
this is less than clear for someone who hasn't run it before.
I think I have it running, but now I need to figure out how to access the admin utils if I want to change tick rate
oh, it's docker compose exec screeps cli
also, note that pushing anything while you have the inspector open just hangs the runner completely
Yeah, I don't think I'm gonna be able to help much there. The more I dig and try to fix it, the more weird errors I get from stuff not looking like it's been initialized
(from the plan, I mean)
fairly sure this isnt strictly private servers being unable to while MMO is, fairly sure i've seen stack errors on private servers as well. in general it seems that screeps seems to (very randomly) sometimes not be able to output error traces, or they end up really strange. no clue as to what may cause them, whether it's a server- or player-side issue, etc.
oh, sure, but this isn't random. a trivial example of infinite recursion always gives a useful error on World, and always crashes the runner on Sandbox
well, i was fairly sure i saw stack traces of infinite recursion on private server, but maybe i'm just imagining things
which server? maybe someone knows somethign we don't about server config
that would be my local one then, which uses alina's screeps launcher. not sure if there is a difference in behaviour with jomik's there
Oooo, dis cool
@haughty cloak Are you unclaiming for gcl?
Not intentionally. But your bot is helping with that a little: took out an rcl8 and another rcl 3 it was using to trampoline elsewhere.
lol
I uploaded a copy of my MMO bot and it still had you as Ally, which contributed maybe
I saw your bot is raising 2-3 low rcls so was curious why
Yeah. I have configs that are different for different server(shard)
Also in the defense are wasteful mode of MMO so no walls
I haven't checked on it in a couple of days (traveling for family) so not entirely sure what it's up to
From your message a couple weeks back, should I be setting something in the "stats prefix" field of grafana config? It was blank. And "stats path" was also blank, pulling 4500 metrics. I replaced "stats path" with "stats" (my super-basic stats are in Memory.stats) and now it pulls 4 stats.
Yep, that's good
(plus presumably the stuff Admin utils pulls down.) That gives me a series like
data
jayseegames_com
shard
adminUtilsStats
users
gcl
Stats prefix is something you can hardcode into the grafana stats path thing. Stats path is a Memory subkey to limit what's being fetched
(is stats prefix pre-pended onto my Memory.stats keys?)
You'll have to update your dashboard to remove the extra stats (from the new "stats path") but that's about it
It's appended
You should get… data.$server.$shard.$optionalPrefix, then whatever is getting pulled
I'm a bit lost as to what you meant by "data.$server.$shard.*"
So, here's an example from my config:
Oh, I see. The data in grafana gets data.$server.$shard as the first part, then $optionalPrefix is the "stats prefix" of the config, then .whatever from the actual Memory.stats block?
Pandascreeps is hardcoded to provide the first 3 segments. Traxus is my Stats prefix. colonies is one of the keys of my Memory.stats object
yep, exactly
Thanks, updated to include Kalgen as "stats prefix" and will wait for it to populate. Much appreciate the help!
I'm being bullied again 😭
😮
Me too! Harassers in the remotes!!
It seems that there is an annoying player in the sandbox🤔
@scarlet fossil it looks like server is unhappy about something, 3.5s ticks
Sandbox still paused or am I being dumb somehow
i think its just paused sadly, i would love to get some debugging on the long term stuff for my rewrite
oh yeah, I never looked at that lol
Could always nuke and restart if figuring out how it's broken is too hard.
I guess, yeah. I mean, it's not hard, it's just… a hassle, as I have to grab the list of everyone active, suspend everyone, then slowly reactivate each player, letting the server tick a few times, then hope I catch the one that's causing the crash
I think with a bit of elbow grease, there's a possibility to script all of it, but it feels like such a niche, especially since that talk with o4 about that where he showed me the vanilla pserver doesn't crash like that
so it turns into an even deeper why does that happen on the community servers and not the vanilla one
I meant if the hassle is too much, delete the DB and wipe. I get that's a hassle too, and doesn't prevent it from happening again, tho.
Yeah. Always welcome entire reset
I'd much prefer a reset over no ticks
It's ticking again!
Do note that it's running 4.2.21, so console.logUnsafe is there
Yay
🎉
i wonder how many months its gonna take for E19N12 to be RCL 8
http://jayseegames.localhost:8080/(http://jayseegames.com:21025)/#!/room/shard0/E19N12
@storm hazel You're having bucket issues here btw
5-10 mins and i'll take a look
sure, no rush. It's just that you're showing up in the logs so I thought I'd give you a heads up since the server was paused for so long
btw, it's good to hear that sandbox is running again, kind of fixed bucket issue
And thanks for notifying
yes
and I think so? I didn't exactly check, tbh, I just wanted to see if it would survive being moved from musl to glibc, and it did
except for the history mod, which I needed to reinstall because it has a binary sqlite installed
can an admin help me please? I signed up to the server with username & email and its not sending me the join up email. So I then tried to re-register using steam oauth, and now im getting "username & email already exists" :/
sure! Email isn't set up in the thing
so make sense you'd get nothing
Can you delete my account and I can start again?
yeah, let me just have a little fight with mongo
let's do that in DMs?
It's a bloody jasper graveyard round here 🤣
yeah, right~
nah, I think that's normal
the pserver's idea of stats is just…not actually there
actually, it's more of a client problem in that case, but it's a bit of a hassle to fix up, and doesn't really bring much
Kalgen with the shard1.clear() signature lol
meh, didn't bother to customize it. also launched the bot with mmo ally list which was "fun" when Harabi's bot did not. 🙂
Uhh, @slender compass, why would you be on the server twice? You have two users in there…
that's what breaking the profile page, since it appears you're using the later one, but the page fetches the earlier one
I just deleted the other one
I accidentally created the first one and spawned in, but then realized i had the other name hardcoded in my bot... so i just created a new account with the correct capitalization
Despawned from sandbox so there is lots of free space!
but then realized i had the other name hardcoded in my bot
Doing this more than once is why my "is this me" function now does a bunch of normalization to account for how I might have screwed up my name XD
anyone else seeing something BIG wrong with this room: http://jayseegames.localhost:8080/(http://jayseegames.com:21025)/#!/room/shard0/W8S9
4 srcs, 2 controllers
"If your bot can figure it out, it's earned the bonus sources" lmao
Oh god would it require twice the upgrading though
How can you grab the second controller
room.controller will give you one...
find(FIND_STRUCTURES).filter
Maybe you've invented the concept of a future season. Multiple controllers that double/triple the room sources
Yeah that'd be the best/only way
whooop
all done
I'm guessing… race condition when recklessly regenerating the same room over and over?
That seems like a silly thing not to guard against, but certainly possible given how....quirky I know private servers are
like, I ended up clickety-clicking a bit too much on that room, and it got stuck with the object half of the last room not deleted while it added the terrain + objs of the next one
Wild
is it possible to reset your password, for code upload, without knowing it? i can log in with steam and access the server, but im on a new pc and cant remember my old password
iirc if you're logged in you dont need to enter your old password
If you used a code upload tool to the server on your old PC you might have the password in the config file for that. That's how it would work for my workflow but IDK if it would for you.
if you can login with steam, just change password after connecting :p ?
Oh no, Harabi is back 😨
Also, I might be able to reset a password manually. I'm not sure it's possible to do a password reset like that, even with Steam auth giving it a go
authmod works exactly like that and i'm quite sure i did it a few times
…wait, isn't there supposed to be a setPassword CLI command?
found it, it's not in the usual "namespaced" stuff, just setPassword(username, pass)
the hidden auth page? really? interesting… I was looking at the in-server Change Password, and this one also asks at least has a field for the old password…
check your DMs for your new old password
you dont have to provide it i think
at least not in steam client
oh, steam client. yeah that might do some Steam Auth magic
Yeah I'm not in the same town as my normal PC until season starts. So I don't have access to my servers.json. I didn't include it in my git repo...."for security" 😂
Heh. Secure from any shady people who might have access to your git repo... like yourself. :P
Yeah exactly 🤔...tbh it's just a habit. It's common practice at work to leave all credentials out of git repos for security.
Yeah, my credentials don't get saved in mine either. It defaulted to that and I never changed it. I always carry a USB with me that has my V1 bot's code and config files on it though. 😂
Yeah that's cool. Generally GitHub is very secure. I just need to start using a password manager so I can have my credentials where ever I need them
I do use a password manager, but I never really thought of putting my screeps code credentials in it. I probably should.
For my own private server all the credentials are so basic that I can easily remember them. My API key for pushing to official servers is much more important and secure though.
keepass with db file on pendrive + dropbox/drive works perfectly for me
I tattoo them on my arm
I use Keepass as well.
I'm running out of arm tho. also, it's costly
I host the keepass db on my web host.
Did it cost you an arm and a leg?
Had to have a skin graft to clear the ones that were leaked
I shut up Monday 🤣
I'm teaching the Jomik image to take a JSON object of Screeps components to git URLs to be able to install random branches instead of waiting for an update to land
What's the pros and cons of jomik compared to original launcher?
what do you mean by original?
if you mean the actual, dev sanctioned setup, the issue is the old Node 10 version and requiring python2
I meant screepers-screeps-launcher?
if you mean Alina's launcher, then it's a bit more… muddy. The main difference between both lies in how they build their images. From a purely… ideological standpoint, a Docker image is supposed to be static. You build it, it has all the pieces backed in, so when you run it the server's ready to go. Alina's launcher doesn't do that, it has to perform the whole server install each time you run it, while Jomik's image has the server pre-setup
It has a big impact in the startup times, mostly.
I see... so you think jomik's is better? I should try to migrate
On the other hand, Alina's image, while a bit more… brittle (because of the "inline install" setup) has more belts and whistles, like an integrated backup tool. Also, most of it is written in Go.
Brittle because if anything ever changes in the npm package dependency list and breaks, then the image is broken. Jomik's, having the whole setup baked into the image, would just keep running, until a new image was published (which would honestly never happen, but still)
It's more of an ideological difference, so… up to you
I find Jomik's easier to hack on because of the non-Go-ness.
I see. I used screepers because of the backup / restore functionality, but I felt those are too unstable...
Sometimes entire setup just broke
Yeah, I don't have any experience on that front, since I just wipe it all clean and recreate any time I need a test server
Yeah. Seems that that's safe option
Would you guys be okay with dogfooding some of the server changes I have scheduled to make sure they're all nice and proper? Now that I have the ability to load up any random server component into the image, we could try and see if the isActive fix, or the extended notifyWhenAttacked stuff actually functions (though the latter would need emails to be set up, I guess…)
I'd be fine with it
Is it even possible to get email from sandbox?
https://github.com/ScreepsMods/screepsmod-mail-aws-ses is the one mod I'm aware of that reenables email notifications
i'm pretty sure it doesnt do the full setup routine (anymore). it does some checks, but then it just boots normally
oh? well then, maybe that has changed at some point. last I remember it was doing the whole isolate-vm compilation while starting up
A not so fun suprise 😄 Few days ago: one room, today: quarter of the map is yellow again
yeaaaah
i think it did some 3+ years back, yeah
Hi! Today's doomsday. I'm gonna upgrade the server to the beta version to see how it reacts to being on Node 24, if there's any breakage and stuff
Cool!
we'll see how it goes
I'm gonna take notes about what and how to back up
maybe at some point I'll write the Jomik support for that 😅
$ node -v
v24.14.0
🫡
ah, right, the broken cli access
@teal sapphire did you happen to have a look at that by chance?
I mean, I can work around it using netcat, just wondering if you were able to confirm if it's a beta problem or something else
4208169 was the first Node 24 tick
hmm, so it's up, it's ticking, but web socket room subscription appear to be broken, so room view doesn't work
oh hell nah
Harabiiiii! ✊
jokes aside, it's something I broke in Steamless. v1.17.0 works fine
Nice to know I'm not the weakest target around Harabi 😄
oh far from it, yeah

I really haven't spent enough time trying to bring OM's combat capability up to something worth
My RCL 7's did get the treatment just as they were about to promote
Somehow my loves your bot than others
Love? I'm being bullied in there
I always feel like my bot has special feeling toward your bot and gadjung's
I can send you energy if that helps 😄
not sure it might
I can stop it... but when I get home
nah, it's okay
I'll send some as an experiment
I mean, you're still spreading a bit too much for my comfort, but
you know I could just… cheat anything I want in, right?
Haha
I could blink those quads out of existence
Can you reset my gcl too?
yeah
Can I?
get a gcl reset? yeah
How? Couldn't find delete id or so
it's a property on the user
From watching sandbox I think it's clear I need to write automated expansion code, I just haven't figured out how to score rooms the best
money: 21394760354.993164,
gcl: 888671402,
oddly enough, you somehow have fractional monies in there
I don't think there's a "best way"
Yeah I have scoring system but I don't believe mine... so I'm using manual direction in season
Yeah, I think I have a vague understanding of what I'm looking for personally but it's hard to quantify
I'm doing what feels like it would be a convolution of sorts? Mostly it's:
- for a given room:
- grab all rooms range 5 of it, if any of them is owned by an enemy, stop
- grab all rooms range 3 (or 2), if any of them is owned by me, stop
- if the room has no controller, stop
- otherwise, get all rooms range 2, check if we have source intel on those.
- if we don't, stop
- calculate pathing distance from each source to the bunker's assumed storage location (means you'd have to actually planned the room at some point)
- sum it all up with magic math
- give a small boost if the room's mineral would balance out the empire's current mineral availability
- etc.
"magic math" is because I can't remember what it's actually doing. It's definitely doing something about walk distance to sources, skipping those that are further than a set number of steps
Hmm. For me the hardest step would be ensuring I have scouted every room in range 5 of the room
My only deterministic scouting is observers
Maybe I build scouting code first
yeah, that's kinda the point I'm at, trying to make the scouting passably more useful so it autocollects the intel it needs
Rule 1 gotta go. Half the fun of auto-expansion is watching it faceclaim your neighbor
oh yeah, put that under an aggressiveness knob~
The "Jasper factor"
Appearing out of the 4th dimension to set up camp on your lawn
I had cleared this room and the next day Jasper was already rcl 3 in there
Checks out
"It's my lawn now" 👴
@past ice Testing suppression toward your bot in sandbox. It might be a good test case for you too
Nice, I'll watch what happens
(Remote suppression)
My bot on sandbox is a bit out of date, it seems to have old code for remotes and likely their defences judging from the roads not being rebuilt
Clever going around, I dont think my bot can respond to a roaming threat like that
I find it funny nobody's having any issues with Node 24. That's a good sign
anyone tried switching their bot's transpiling target to that?
Did they roam well?
working great with my rust bot atleast! no MVP wasm or sign-ext lowering needed anymore 😄
.. definitely good reminder to fix my resource allocation issues tho, haven't taken a look in a while to how my bot is doing in sandbox
just for some ideas, how does everyone here handle space allocation? some hardcoded limit per room or possibly a dynamic global limit?
currently im trying to go for something akin to "2% of all storage space is used for this resource type" but that feels a bit wasteful, allocating space for resources i won't even have for a long time
Your suppression seems to be effective, at least against my bot
I think I found the stack overflow 😅 ```ts
remove(force?: boolean): OK | ERR_BUSY {
const ret = this.remove(force);
if (ret === OK) {
Overmind.combatIntel.unregisterThreat(this);
}
return ret;
}
funnily enough that didn't even murder my CPU. Bot kept trucking around at 9999 bucket
aww, Harabi's coming for me again 😭
is there some claim/gcl limiting in sandbox going on?
been fixing lots of bugs today in my bot and stumbled upon my bot constantly trying to auto-expand but never succeeding
it think it has 22 GCL, 21 rooms claimed which lines up with what i see in my profile and sends a claimer & colonisation squad but always gets the ERR_GCL_NOT_ENOUGH result
http://jayseegames.localhost:8080/(http://jayseegames.com:21025)/#!/history/shard0/E12N8?t=4936909
i really don't see anything that could be wrong in my code that could cause this, controller is freshly gotten a couple lines above
match creep.creep.claim_controller(&controller) {
Err(ErrorCode::NotOwner | ErrorCode::InvalidTarget) => {
creep.creep.attack_controller(&controller).should_be_ok_dbg((creep, controller), "claimer attack controller because fail to claim");
},
Err(ErrorCode::GclNotEnough) => {
let _ = creep.creep.say("no gcl! D:", true);
self.claimer_waiting_for_gcl = true;
},
e => {
if e.should_be_ok_dbg(creep, "claimer claim controller") {
rdm.force_update_next_tick(controller_pos.room_name());
}
}
};
ohh, i figured it out
for some reason the game thinks i've claimed E13S5 when i haven't
so in the claimController check it sees 22 rooms (all my rooms + E13S5) so it thinks i've reached my cap
lemme check…
…right, broken CLI on Node 24
that's gonna be fun
that's your list of rooms: ```
rooms: [
'E3N12', 'E4N1', 'E5S2',
'E13S8', 'E12S5', 'E11N12',
'E16S8', 'E15N19', 'E9N6',
'E4N2', 'W19N11', 'E11N19',
'W19N8', 'E13S5', 'W17N14',
'E7S14', 'W15N2', 'W4N19',
'W1N18', 'E9N15', 'E7N2',
'E1S8'
],
and yeah, definitely a core there:
i guess that was just a client bug and nothing related to the room list being weird
unlucky timing
E13S5 is indeed in that list, but i've not claimed that room
i wonder how that happened
I mean… there's a spawn there
do spawns just count??
no, but I mean… you have to have owned the room at one point for a spawn to get placed
some infra from the nearby room
ah, yep
its been there since forever i guess
seems like the rooms i own is a list that just gets incrementally updated whenever controller ownership changes & doesn't check the game state
i guess there is a path that somehow avoids the removeRoomFromUser call and still removes my ownership of the controller
earliest tick still in history: http://jayseegames.localhost:8080/(http://jayseegames.com:21025)/#!/history/shard0/E13S5?t=2470000
oddly enough, anything before that is a 500 in the console, but the server still serves an empty JSON for it
heh, history doesn't stop behaving oddly i guess
Maybe… I mean… it might make sense to return a 500 to make the client not bother trying to make sense of an empty chunk
still, no idea what's the story behind that extra spawn, and why the game didn't fix that up when you lost control of that room
structures don't get removed when you unclaim tho?
i guess my bot abandoned the room at some point
so… 21 owned rooms at the moment, right?
yep
[
'W4N19', 'E9N15', 'W17N14',
'E3N12', 'W19N11', 'W19N8',
'E9N6', 'W15N2', 'E7N2',
'E5S2', 'E12S5', 'E1S8',
'E7S14', 'E15N19', 'E11N19',
'W1N18', 'E11N12', 'E4N2',
'E4N1', 'E13S8', 'E16S8'
]
gah, not having readline caps is a pain
perfect!
even if its an empty array it should just add all rooms i own the next tick to it
I mean… we can test that 😅
worst case scenario i will get 43 rooms!
sounds like a great deal to me 😄
http://jayseegames.localhost:8080/(http://jayseegames.com:21025)/#!/room/shard0/W5S7
instantly claimed room #22, excellent!
hmm, it rewrote the rooms array
mh?
rooms: [
'E15N19', 'E5S2', 'E7N2',
'W19N8', 'E11N19', 'W15N2',
'E16S8', 'W17N14', 'E7S14',
'W19N11', 'W1N18', 'E4N2',
'E11N12', 'E1S8', 'E9N15',
'E13S8', 'E4N1', 'E9N6',
'E3N12', 'E12S5', 'W4N19',
'W5S7'
],
ahh
ah, right, it claimed and recalculated that, with W5S7 as the last one
thank you very much for the quick fix! 😄
now to.. fix that bug.. 0% chance of surviving
yeah, that's… an interesting situation you got there
Not sure what happened. Both the claim intent and the controller tick intent have calls to addRoomToUser/removeRoomFromUser
so that should definitely not have stayed in the list
yeah, i've got no clue how it got into that broken state
@teal sapphire sorry for the ping, but… given this is odd, maybe you'd like to know?
this is on Node 24
mhh, how big is my room list now btw?
mostly the question of does it include W15N2?
rooms: [
'E15N19', 'E5S2', 'E7N2',
'W19N8', 'E11N19', 'E16S8',
'W17N14', 'E7S14', 'W19N11',
'W1N18', 'E4N2', 'E11N12',
'E1S8', 'E9N15', 'E13S8',
'E4N1', 'E9N6', 'E3N12',
'E12S5', 'W4N19', 'W5S7'
],
it doesn't
okay, thats good
Any way to reproduce that?
not exactly sure, since I don't have the history of what happened
my guess would be (but I'm not sure it makes sense) that an invader popped up right as the room downgraded or something?
Can't perform a deep investigation right now, sadly
Please ping me if you're happen to figure out the reproduction
sure thing! I'll do some check to see if there are other inconsistencies like that
i really can't find any leads in the code as to why this could have happened
downgrading either by user or invader manipulate the downgrade timer which gets checked in controller ticking which is sane
can't see anything wrong with unclaiming logic either
best idea i've got so far is maybe 2 people attempted to claim the room in the same tick, both getting addRoomToUser called but only the last person actually getting the claim but im not sure if thats possible
either via 2 .claimController calls or possible someone spawning into that room on the exact same tick
but Jasper's pretty much the most intense claimer I've seen, sooo…
yeah, I don't think it's a bot-side thing, this sounds more like the server getting something mixed up
absolutely
I mean, that while loop looked weird to me until I remembered that a room might not tick for a while, and this is just handling progressively downgrading it for all the time it didn't get to tick
can't wait to be on the end of that email-storm that causes lol
i do think there is a race condition between the claimController intent and someone spawning into that exact same room on the same tick, but that would be insanely hard to pull off and i doubt that has happened in this case
actually no, nevermind
it can do funky stuff but not that
yeah, I see it's checking the user before doing the whole place-spawn dance
yep, but its chained with .then() calls which might give the processor time to do the claimController intent, hence why i thought a race condition might be possible
though it doesn't update said user's room list until the controller ticks
but in the actual line where it updates the controller it specifically looks for a controller at level 0 to claim, which is a atomic check & update so no race condition
im out of threads to pull i guess, imma just chalk it up to magic bitflip or something idk
ah, right, there's a small window where "place-spawn checks, is green, schedules more stuff, while claim controller swoops in, then gets blasted off by place-spawn finishing"?
yep
so, I got everyone's current rooms list, but checking that against the real list given the broken console sounds like my idea of un-fun
27 users
fun fact:
while in this endless auto-expansion broken state my bot has managed to attempt to expand 30+ times to the same room... for 30 different rooms
in total 2985 auto-expansion attempts where made
I pushed an update to my bot that includes rudimentary automation, so if a blue triangle is being a bother let me know
Heads up that the server is now running custom server components (I'm testing them out 😅).
So a bunch of the in-flight fixes against the game's repo are in there, namely:
- isActive checks should be free-ish
- creep renewal has a list of extensions to use
- the attack notifications stuff is there, so there's a spawn flag, an accessor, etc.
- CostMatrix.clone() fix
- unspawned powercreeps TTL fix
- powercreep name limit is 100 everywhere now
- the weird pull on death bug someone fixed is there too
and I'm testing out a change to how history gets saved (moved it at the start of the tick, before anything, and saving the data for the $tick-1 tick instead of in the middle), that might help with replays looking weird
Is the server down or just me problem? @scarlet fossil
yeah, it's undergoing some maintenance
I'm test-running the Mongo mod update, so everything can be updated to a recent mongo & redis version
just hit a couple of snags where it's now consistently wiping the db on start-up, even though the data is there from the restore
Nice, gl with the snags
Okay, server back. If you spot anything weird, do tell. It's running 4 Mongo versions above what it initially runs
auch also 4.5-5s ticks here 🙈
yeah, I rolled it back to Mongo 4 😩
thank god my backup script works
history might be weird though
fixed history by just yanking everything after the backup
The terrain seems a bit odd on the world map, deleting cache didn't seem to help
this one? ```js
dbs = await indexedDB.databases();
dbs.forEach(d => indexedDB.deleteDatabase(d.name));
That did fix it
weird… you somehow got the alpha map for something else in there
I did do a server side refresh of it all, for the normal map
Hmm. It is on steamless and I only really check mmo and sandbox, I dont think they should be intermingling
yeah, especially if you have the subdomain thing of steamless
what might have happened is that you ended up looking at the server while I was in the middle of the restoration crisis
so you saw a random newly generated map
Ah yeah I did load it yesterday when I pinged you
It didn't load anything on the UI but might have loaded the map somehow
yeah, that might have been it then