#random (2024-04)

I set up openstack for a client team and my findings are that it just is too much overhead and docker-configuration, awslocal configuration, a collection of shell scripts etc

it didn’t make sense for a team of ~10 engineers but still they wanted a full local loop of their lambda flow which involved things like s3 bucket notifications for triggering lambda etc

I think localstack is good if you built with incorrect patterns and don’t have a good way to locally execute your code. But in an ideal world you should not need localstack and should have a strong local dev flow and from there move to a dev cloud environment.

So learning it and trying to provide it as service will not be beneficial for the long run ?! my main interest was finding a way to help mid-companies achieve their on-premise plans outside the cloud providers options(AWS outpost..)

I wouldn’t say that it’s an extremely valuable skill to have. I haven’t seen many organizations that are using it at scale, and with proper configuration of localstack it’s basically the same skillset as AWS. I would focus more on sharpening pure-AWS skills

Understood, thanks for your feedback

Yea, I think if the purpose is testing locally, local stack is where things are headed. Much simpler. Open Stack is overkill unless the objective is to create a self-managed cloud on bare metal.

ohh @forswearbeetle i’m sorry I completely swapped openstack and localstack in my head when I read your first message

And what I meant to say is that I set up localstack. I do not have experience setting up openstack

Incidentally, a new startup out of ycombinator is trying to provide a new alternative to open stack (with a hosted version as well) https://www.ubicloud.com/

https://www.ycombinator.com/companies/ubicloud

I have set up Openstack in a test environment around 2013, it is a good project to learn but not for making a living or commercial project, many big companies had failed on it

@Erik Osterman (Cloud Posse) I think cloudstack is a true alternative to openstack for now

@Hao Wang what do you see as an alternative and open source to learn ?

also used opennebula

Heh, didn’t think cloudstack still existed. I only did a small PoC 12+ years ago when was at CBS.

Interesting. So I used it in 2011, just before it was acquired. Didn’t realize Citrix donated to Apache Foundation.

not that much of articles and blogs around it but the documentation seems decent, easier than openstack tbh

Yes, when I tried cloudstack it was pretty easy to get up and running.

I did a lot of openstack back in the day. We would use cobbler and pxe to bring up dozens of compute for clients and some wanted their own clusters. In the end, the tech is mostly about learning rabbitmq… I feel like that’s the only long term skill I got out of it. And learning how beam/rabbit work and learning some erlang/elixir would need to be on your skill todo’s first.

Customers leaned on it as an alternative to ec2, and we would basically find value in it because customers were buying hardware and we would make sure it sliced its hardware neatly into clusters.

That’s still something that can be tough to do with kubernetes and other tech, but to be clear, it’s going to move you closer to hardware

Getting closer to hardware is literally the opposite of getting closer to cloud. Not that they can’t work together, but I would not consider them overlapping or even reinforcing cross-cuts of the same problem

tinkerbell is another interesting project

based on containers

forwarded this video from #security, from now on, for all the open-source projects, it is expected that the team of maintainer or co-maintainer will be audited internally and externally in all the open-source projects, which is a good thing

you never know who is behind them

hope he can come back to DevOps community or his own company as Jobs did

MIT license is sad

I would listen to this on repeat!

IMO, MIT is pretty happy as far as licenses go. It would be hilarious to do this based on a source-available license that Redis, HashiCorp, ElasticSearch, etc have adopted

optimistic to see the doors are closing to those products

amazing nature

that view almost makes the earth look like an eyeball

Right? It’s almost creepy looking

(I forgot that sun acquired mysql, before oracle acquired sun)

I worked for SUN then so got some impressions still, SUN failed its new CPU and then fell hard

chatgpt 3.5 is used

gemini free is used

Looks like 4 is getting some degree of updates

Looks interesting. I think for many of us, there’s too many new cool tools like yours we would like to try out, but not enough time. Some examples / tutorials / videos allowing me to learn at least some cool bits of the tool in minutes rather than hours would be cool

there’s too many new cool tools like yours we would like to try out, isn’t that the truth!

especially every time after office hours

YES @Pawel Rein I’m working on that

I tried demoing some cool bits in this 2 min video, If you think it’s interesting enough you can give it a try here https://histudio.ai

Really appreciate the candid feedback !

https://atmos.tools/schemas/atmos/atmos-manifest/1.0/atmos-manifest.json this?

Yeah, thx

They don’t do GH donations (or any it seems) – Crazy.

They have charm cloud: https://charm.sh/cloud/

but mostly i think they make money through paid support

That’s awesome. Again… how does that get enough traction to be profitable is my question, but I love it.

Maybe atmos v3 needs to be built on Charm

I want to do a lot more. Just we need more sponsorships!!

now I got used to Azure after a year

Honestly I don’t hate the concept of resource groups.

Yes! I started with AWS and then started working with Azure. I really appreciate the way ALL the resources for a project are encapsulated in a resource group. In AWS you can tag but it still leads to resources getting overlooked or left behind when projects get decommed or renamed. Especially back in the click-ops days.

Interesting idea, but I don’t want AI anywhere near my terminal. Just feels like a security nightmare waiting to happen.

I also skip over everything related to AI on hacker news and reddit, so I’m clearly not the target audience.

this actually looks pretty cool to me, since I’m already a heavy vim user and havent ever really been able to transition over to VSCode. Plus rendering markdown and images is nice to have. Downloaded and giving it a chance!

Let me know how it goes! I’m hesitant to switch to something too quickly as I just got on Warp last year and like it, so I’d like to hear first-hand experience.

@Matt Gowie do you have any favorite Warp features you use? I just made the transition to it

Honestly I’m coming from iTerm2 and my feeling towards Warp is that the Emacs key combinations that got ingrained in me from 2011 just work better. Things like alt+shift+b for back selecting a word and what not. I think my bar was kinda low. They have better history search and tab completion too.

But I’m definitely not a power user of it. I haven’t really used their AI stuff at all.

Agreed, I haven’t used any of the team or AI features, but the tab completion and search is very nice!

I tried out wave for a day, but honestly didnt love it. We heavily use docker shells at Cloud Posse, and wave makes it pretty awkward to handle. Kind of a hard stop for me

Huh – good to know. What’s it doing in your screenshot? Separating the execution from history or something?

yeah it creates a separate session for the docker shell, but continues new commands in the original shell below. You can click and enter the docker shell if you want, but it’s a lot of clutter

Weird. Probably should be an issue for them cause that sounds like a bug. They must know about it though cause entering a subshell is a normal thing…

Holy cow, Warp collects more telemetry than the NSA. I tried it out this week and opted out of all optional telemetry, but after blacklisting the domain so it wouldn’t call back to app.warp.dev, I realized how much it’s sending outbound. https://docs.warp.dev/getting-started/privacy#exhaustive-telemetry-table

I have often said that documentation should be executable

You’re making this a reality

Thanks for the feedback, would love to know what else you want from the project to really make it accomplish those goals in real life use-cases!

can you provide a brief info on the product? post a link to demo here as well

Definitely. https://www.glueops.dev/ https://www.youtube.com/watch?v=QHtY2jOV1M0

GlueOps is a PaaS that makes it easy for you to deploy apps by combining the best open-source tools into one platform, similar to Heroku/App Runner/AppEngine/Vercel/Netlify. You don’t have to worry about the technical details or maintenance of the OSS tools.

While we use advanced technologies like Kubernetes, you won’t have to use any complex commands. Instead, you just interact with a simple git repository to manage your setup, and we take care of the rest.

Looks cool. Is it K8 only, or can you also make the ephemeral environments for e.g. AWS Lambda API’s?

Thanks for the feedback and questions. Much appreciated. We have the ability to support serverless type apps but don’t have any customers running those today. As for how we do it, we just use https://github.com/kedacore/keda as our event based scaler. For our current customers that were using serverless/functions they were using it like a cronjob and we just migrated them to a “job” within the context of kubernetes.

If you do find something that works well I would love to hear about it!

I have seen a few over the years. The problem is they always end up shutting down. We manage our own, but I would rather not. But I am scared to switch only for the service to shutdown.

Telegram is free forever! And now has group chats with channels. None of the stupid tenant-scoped users crap and cross-server channels, just DM people wherever

https://www.cnbc.com/2021/12/09/cloud-software-maker-hashicorp-hcp-starts-trading-on-nasdaq.html?ref=runtime.news

which means Hashicorp is eager to sell itself, otherwise…

with any means it can use including license change, which is the last one but quick to please investors

As a fool that bought at the IPO, I can tell you that I’m still in the hole. Even after the acquisition announcement

Lol the acquisition is when the funniest thing begins

after pandemic, money is not easy to get, big financial institutes are in hard time, and pass the toughness down to the listed companies

Yea, so this is the part that’s a lingering mystery for me.

There’s a theory that the BUSL change was a requirement for the acquisition to proceed, so that IBM wouldn’t need to tarnish it’s reputation by making the license change

And if that’s true, what’s the point of this fork? And the fact it’s in the LF, means it’s got some legs.

back then (4 months ago), the intense is not disclosed to IBM employees yet, and the IBM employee did the hard fork along with Linux Foundation

and now community got concerned about the OpenBao to be poisoned pill since the IBM employees got interest conflict

oh shit! I didn’t think about that angle - that it could be a poison pill. Any posts on that?

saw some discussions, the pill may not be by purpose but just the timeline does show the acquisition is not planned well even with strong investor supports

forking a rug-pull repo is saving your next generation, to give them a job opportunity not being controlled by someone behind the scene

imaging me working in a state-owned company for the whole life, instead of talking to you guys here (my Mom said many time her life is wasted)

However, the system is currently experiencing maintenance limitations due to faulty quick disconnects causing water spray. Given the expense and downtime associated with rectifying this issue in the last six months of operation, it’s deemed more detrimental than the anticipated failure rate of compute nodes. Approximately 1% of nodes experienced failure during this period, primarily attributed to DIMMs with ECC errors, which will remain unrepaired. Additionally, the system will undergo coolant drainage.

At least they’re honest. Sounds like a money pit!