IRC logs for #openttd on OFTC at 2024-07-10
⏴ go to previous day
00:38:35 *** Wormnest has quit IRC (Ping timeout: 480 seconds)
01:26:46 *** Wormnest has joined #openttd
02:07:40 *** Wormnest has quit IRC (Quit: Leaving)
02:45:01 *** D-HUND has quit IRC (Ping timeout: 480 seconds)
08:24:21 *** mindlesstux has joined #openttd
08:47:30 *** jinks has quit IRC (Remote host closed the connection)
09:31:14 <LordAro> needs ranges, clearly
09:43:53 <peter1139> Hmm, so I stick my power meter on my heavily laden touring bike or not...
09:51:27 <peter1139> Why not? Hmm, well... not sure.
09:52:36 <peter1139> Moving them from bike to bike means the threads get worn a bit, but also the threads are less likely to get seized... I dunno.
09:53:13 <peter1139> Might help ensure I don't go too hard on the knee? Excuses...
09:53:40 <peter1139> Oof, it's raining :(
10:05:58 <peter1139> Did not expect to be packing a jumper and waterproofs...
10:06:58 <LordAro> if the weather could improve drastically for 3 weeks time, that'd be great
10:14:32 *** mindlesstux has quit IRC (Ping timeout: 480 seconds)
10:19:15 *** mindlesstux has joined #openttd
10:35:32 <yiffgirl> man when you only need to recompile one or a few objects instead of the whole thing from scratch it's orders of magnitude faster
10:55:31 <yiffgirl> one of these is current master with tree placement set to original. one of these is current master with tree placement set to improved. can you tell which is which?
11:50:04 <peter1139> I had a patch (stash) for that but couldn't find it, so I rewrote it again :p
11:56:22 <merni> peter1139: Imagine a "peter1139 Patchpack" with all your patches somehow smashed together :p
11:58:49 <LordAro> we generally call it 'master'
12:00:21 <merni> no I mean including the ones which haven't been PRd/merged lol
13:29:17 <peter1139> Okay, well I think that settles it.
13:29:25 <peter1139> I simply cannot get the touring bike in my car.
13:30:13 <peter1139> I would have to remove the front wheel (easy) and the front mudguard (not so easy)
13:44:04 <peter1139> (And maybe the front rack (also not easy))
14:35:08 *** DorpsGek has joined #openttd
14:35:08 *** ChanServ sets mode: +o DorpsGek
14:37:01 <truebrain> right ... seems DorpsGek froze up .. odd
15:05:47 <truebrain> bit sad, if you ask me
15:08:14 <peter1139> Hmm, so `std::vector<std::shared_ptr<X>>` vs `std::vector<X>`.
15:08:45 <peter1139> The former is cheaper to sort. But pointers.
15:09:03 <truebrain> sort-by-pointer-value? 😄
15:09:40 <peter1139> You can of course just move the pointer instead of the actual object.
15:10:47 <_jgr_> Using shared_pointer implies that you need reference stability and for the items to outlive the vector, just using the value implies that you don't need either of those
15:23:25 *** HerzogDeXtEr has joined #openttd
15:26:42 *** nielsm has quit IRC (Ping timeout: 480 seconds)
15:27:34 *** Wormnest has joined #openttd
15:41:21 <peter1139> Hmm, shame, there is at least one place where the list is copied, and also where the pointers are used after being inserted into a vector.
15:49:20 *** gelignite has joined #openttd
16:23:41 <peter1139> Hmm, maybe a 6.30am start.
16:24:18 *** gelignite has quit IRC (Read error: Connection reset by peer)
16:24:46 <peter1139> Have I been Zorged in the forums?
16:28:14 *** gelignite has joined #openttd
16:36:34 *** Flygon has quit IRC (Read error: Connection reset by peer)
16:44:09 <andythenorth> oh that cdist thread
16:44:18 <andythenorth> I drafted a reply there, but deleted it
16:44:29 <andythenorth> afaik all this stuff about cdist calculating capacity is woo
16:44:36 <andythenorth> I have notes about it from fonsinchen
16:45:00 <andythenorth> it just uses distance for demand, then it does [something] about symmetric
16:45:15 <andythenorth> cdist still misunderstood?
16:46:57 <merni> andythenorth: Maybe you could share those :p (if you haven't already somewhere)
16:48:29 <merni> "saturation of short paths before using high-capacity paths" does make me think somehow capacity is involved but perhaps not
17:36:59 *** efessel has joined #openttd
17:36:59 <efessel> truebrain: FYI - I've moved some compute resources over to Oracle Cloud from AWS. It's nice having a $0 bill every month.
17:37:22 <peter1139> You managed to get a response from them?
17:37:39 <truebrain> in fact, we just created an account at OCI. Now to figure out how to do cloudinit via Instance Pool ... #documentation-sucks-yo
17:37:46 <efessel> So far no big complaints, although their free compute means arm compiles which is no big deal
17:37:59 <truebrain> We already run on ARM, as it was cheaper on AWS too 😛
17:38:15 <truebrain> but glad to hear it is working out for you 🙂
17:38:23 <efessel> I'm using Terraform w/ their provider
17:38:37 <truebrain> yeah, but that is just half of the puzzle 🙂
17:39:06 <efessel> Right, I have't messed with dynamic instances at all
17:39:15 <efessel> terraform runs the cloud init
17:39:17 <truebrain> I like auto-deployments, where everything is installed automatically. And normally you use cloud-init for that. Which is "easy" when spinning up a single instance, but somehow it is hiding from me how to do that with Instance Pooling 😦
17:39:40 <efessel> Yep, I'm a fan of the computer doing all the work too
17:40:04 <truebrain> btw, Pulumi is a great wrapper around Teraform, to make things easier to understand in many ways
17:40:17 <efessel> well terraform doesn't do the cloud init, it's specified in userData prop on the instance config. Not sure how that works with instance pools though.
17:41:08 <truebrain> that is the weird part .. userData exists in Instance, but not in InstanceConfiguration
17:41:16 <truebrain> not sure what that is about .. but it just means I have to actually read 😄
17:42:23 <efessel> yeah I def had to read teh dox for a bunch of things, assigning a "reserved" public IP isnt' a strait forward process
17:43:06 <efessel> I assume you've updated your account to Pay as you go?
17:43:19 <truebrain> you have to; otherwise there won't be any instances to spin up 😄
17:43:38 <efessel> Right, which is odd. What's the point of the trial if you can't "try"
17:43:45 <truebrain> I wrote them as such
17:43:46 <audigex> truebrain: Oracle cloud’s interface is awful, puts me off a lot
17:43:54 <truebrain> and I got a weird reply back, dodging the question
17:44:01 <truebrain> so I am sure I am not the only one who complaint about it 😛
17:44:02 <efessel> Their support is powered by AI
17:44:13 <truebrain> audigex: as with ALL the other Cloud interfaces. None of them are any good, or understandable. None.
17:44:13 <efessel> So you'd be better off with Chat GPT
17:44:31 <efessel> (found that out after a ridiculous back and forth)
17:44:41 <audigex> truebrain: Yeah half the time it feels like they’re just trying to sell training
17:44:41 <audigex> AWS, Azure, Google are all a lot more intuitive but “good” certainly wouldn’t come into it
17:44:48 <truebrain> efessel: At a certain point it told me to learn how to prompt an AI
17:44:50 <truebrain> which made me laugh
17:45:29 <truebrain> this is the world we now live in ... AIs telling us to prompt them better ...
17:46:18 <truebrain> anyway, reason for InstancePool over Instances via Terraform, is that an InstancePool automatically creates a new instance if another one crashes for what-ever reason
17:46:34 <truebrain> self-healing and shit 🙂
17:47:06 <truebrain> `The number of NVMe drives to be used for storage. A single drive has 6.8 TB available.`. Lol.
17:47:43 <efessel> Oh and their block storage restrictions are weird. You *must* use at least 32gb for boot.
17:47:53 <efessel> And if you want to add an additional, minimum is 50gb
17:48:06 <truebrain> lol; well, it is dirt cheap anyway, so I guess it is fine
17:48:55 <efessel> Right, but if you spin up 4 instances, now you need to pay $ for detachable storage (well sure, poo poo for a few bucks a month lol)
17:49:26 <truebrain> from the quick math I did, it is 10 dollar on OCI, and 90 dollar on AWS, for the same hardware 😛
17:49:33 <truebrain> and that is just the instances
17:49:39 <truebrain> so there is some room in between those two, I guess 🙂
17:49:53 <efessel> Yeah hah not complaining at all, just weird to me
17:50:04 <truebrain> I was a bit annoyed the "Free Tier" storage is just for 2 block devices .. means you can't make a cluster of three (a cluster of two isn't a cluster, mind you) for free 😛
17:50:34 <efessel> Well if you can't spin up any instances, the point is moot
17:50:48 <truebrain> if you have the PayAsYouGo, you still have the "Free Tier" stuff 🙂
17:50:59 <truebrain> so on my personal account I run 2 instances for free, but I have PAYG active 🙂
17:51:05 <truebrain> I just can't make it three, as that will cost
17:51:19 <truebrain> not for the compute, but for the block storage 😛
17:51:35 <efessel> huh? you should be able to run 4 ampere for free, no?
17:52:01 <efessel> Thought they give you 200gb block storage for free
17:52:07 <truebrain> over 2 block devices
17:52:29 <truebrain> `Up to 2 block volumes, 200 GB total. Plus 5 volume backups.`
17:52:47 <efessel> hmm I didn't read that, where is that ?
17:53:36 <truebrain> no clue if they actually check for the count, mind you. But it reads on the label you only get 2 free
17:54:52 <efessel> hmmm I don't see it in this one
17:55:44 <efessel> Yes, I see that from your link
17:56:10 <efessel> but that's the only place I've ever seen that. Wonder if it's changed or a typo
17:56:14 <truebrain> haha, owh, the sneaky extra conditions ... "in the home region" ... lolz
17:56:29 <truebrain> and you cannot change your home region after creation, ofc 😛
17:56:54 <efessel> Yeah, and I set my home region after finding out there's no AMD Micro instances available 😦
17:57:43 <truebrain> but if the "Two" is old, that would be nice. Means I can run a cluster of three for free 😄
17:58:10 <efessel> I'm checking my expenses, since I've provisioned three...
17:59:06 <truebrain> also, I guess it means everything will be running in Amsterdam for OpenTTD 😛
17:59:11 <truebrain> as that is the home region, it seems 🙂
17:59:16 <truebrain> not the best location, compute-wise, but what-ever
17:59:43 <efessel> but maybe they don't consider boot volumes in there
17:59:44 <truebrain> (no zone redundancy in Amsterdam; in Frankfurt there are 3 zones)
18:00:34 <efessel> maybe ask their AI support to clarify
18:03:02 <truebrain> `Your account provisioning is in progress.`
18:03:06 <truebrain> that also takes for ever ...
18:03:41 <efessel> Oh yeah - took 8 hours for me I believe.
18:03:55 <efessel> I've read for some people it's taken days
18:04:24 <truebrain> upgrade requested at 12:51 .. so 7 hours in already 🙂
18:05:13 <truebrain> okay, I think the cloud-init can just be done via metadata .. something to try out tomorrow, I guess 😄
18:07:52 <truebrain> Although I would like something a bit easier to manage
18:07:55 <truebrain> but never found any 🙂
18:08:03 <efessel> I thougth Nomad was the easier alternative to K8s 🙂
18:08:24 <truebrain> like with kubernetes you have three issues: the network/storage, the cluster itself, and the deployments within
18:08:28 <truebrain> with nomad you only have the last one
18:08:41 <truebrain> but they have their own language, which is just a bit annoying to quickly do things
18:08:49 <truebrain> as I always forget their syntax, and have to spend N time reading up again
18:08:57 <truebrain> I rather have something I can just send docker-compose files to or what-ever
18:09:05 <truebrain> something a bit more mainstream
18:09:20 <truebrain> (and no, `helm` and `kustomize` both suck balls too 😛 )
18:09:45 <efessel> I agree, docker-compose would be very nice
18:10:01 <truebrain> You now have software like coolify, which is basicaly self-hosting Vercel
18:10:10 <truebrain> which is nice, but not yet cluster-ready in a sane way
18:12:42 <efessel> I'm still using k3s with success. Agree though - helm/kustomize is a pain, but I think I have them managed via terraform and gitops pretty well. Using ArgoCD to manage the state of the cluster
18:13:09 <truebrain> ArgoCD is nice; if you need to use k8s, ArgoCD is the way to cope 🙂
18:13:10 <efessel> k3s includes traefik which works well
18:15:36 <efessel> also it comes with local storage provisioner, but that won't work across nodes
18:15:49 <truebrain> which makes it non-HA, ofc 🙂
18:15:59 <truebrain> haven't checked what OCI has for that
18:16:18 <efessel> They apparently give you one free k8s service
18:18:06 *** gelignite has quit IRC (Read error: Connection reset by peer)
18:20:16 *** gelignite has joined #openttd
18:50:38 <peter1139> Helicopters expiring prevent the game from continuing?
18:51:25 <audigex> peter1139: It’s certainly an inconvenience if you’re using them and suddenly you can’t buy any more
18:59:44 <peter1139> No, it *stops* the game.
19:14:24 <audigex> peter1139: If you’re referring to the last reply on the forum I think that’s just a phrasing issue, they mean the game can’t continue *with helicopters*. It’s clumsy wording but they obviously aren’t saying the game stops
19:20:05 <andythenorth> Is there some football to watch?
19:30:33 <truebrain> efessel: I can't get boot volumes lower than 50GB. How did you do 32?
19:30:51 <efessel> I got the # wrong - I think its 47
19:31:16 <truebrain> 46.6GB it says, but I can't get it below 50 😛
19:31:22 <truebrain> but okay, not 32 😦 Darn ..
19:31:40 <efessel> Boot volume 46.6, but detachable block storage minimum is 50
19:31:52 <truebrain> such a waste of space/money
19:32:30 <truebrain> their NLB is funny. Proxy Support? Only for TCP and IPv4. Not for anything else. Shoo-shoo.
19:32:38 <truebrain> So I need to do the same as on AWS, run my own NLB
19:32:42 <truebrain> at least it is cheaper 😛
19:32:55 <efessel> Yeah, I read up you did a lot of IPv6 right?
19:33:15 <truebrain> all our services are available on both IPv4 and IPv6
19:33:38 <truebrain> and I don't want my logs to show internal Oracle Cloud IPs; that makes a lot of things much harder 😛
19:33:45 <truebrain> so we use Proxy Protocol to still see the original IP
19:33:51 <truebrain> but .. OCI is like: only in 25% of the cases
19:34:31 <truebrain> no biggy, I have all the code ready for it anyway
19:57:50 *** amal[m] has quit IRC (Quit: Client limit exceeded: 20000)
19:58:25 *** FLHerne has quit IRC (Quit: There's a real world out here!)
20:07:48 *** FLHerne has joined #openttd
20:58:34 *** HerzogDeXtEr has quit IRC (Read error: Connection reset by peer)
21:03:15 *** Flygon has quit IRC (Read error: Connection reset by peer)
21:05:07 *** gelignite has quit IRC (Quit: Stay safe!)
21:30:38 *** nielsm has quit IRC (Ping timeout: 480 seconds)
21:44:02 *** tokai|noir has joined #openttd
21:44:02 *** ChanServ sets mode: +v tokai|noir
21:50:48 *** tokai has quit IRC (Ping timeout: 480 seconds)
continue to next day ⏵
