Xlskubectl – a spreadsheet to control your Kubernetes cluster | Hacker News Reader

Xlskubectl – a spreadsheet to control your Kubernetes cluster

github.com

140 pointsjhoelzel25 comments

> Q: Is this production-ready?

> A: We're looking for fundings to take this to the next level. Replacing YAML with spreadsheets has always been our mission as a company, and we will continue to do so.

Sounds like straight out of a comedy movie made for infra-nerds! Made me laugh out loud :)

The project itself is awesome, makes a lot of sense. I'd love to keep all infra-related things in a spreadsheet where I can hook it into what I normally do in my spreadsheets, so really cool idea.

It needs charts and shit. Then you could dump Prometheus too!

Crap I need to do a deploy and switch to the backup provider but I can't because my spreadsheet's internet connection is down!

Minimalism is powerful at conveying truth.

/humor Just don't input the wrong numbers. "Oops pasted a date-formatted cell .. spinning up 19258 instances ..."

It's absurd how similar data is, how generic it is, but how arbitrary interfaces formats protocols and other layers impede cross-functioning. So much resistance, so much impedance!

It fizzled & failed (also got replaced by the web & APIs) but OLE/COM and CORBA, those were interesting historical junctures where we expected computing as a whole to weave together better. All is one!

I love this project. Makes perfect sense. Hell yes I want a sheet for each resource & to be able to edit it.

Just imagine, a spredsheet for every resource definition...

Yes I can see it, I would not want to work with this at all and when management realises all they have to do is play with the spreadsheet, the dream crumbles just as quick as it came to be :D

I hate this.

I can't wait to try it out.

I realize this is all a "bad idea" but I'd like to hear what the root reasons are that it really is such a bad idea.

It certainly seems to be a more maintainable way to do it than many "good ideas" I've thought of.

Do you want a complex scaling algorithm that takes the number of visitors and last month's revenue into account? Just write a simple formula!

What jumps out to me is that spreadsheets, and especially Google sheets, are opaque and often have weird issues that are difficult to diagnose. Eg, someone disabled formatting on a column for some reason, and now your mathematical functions aren't working.

I really don't like the idea of a stateful document, in general but for configuration especially. If I have to look through the edit history to figure out why the document is behaving in a certain way, that's a hard no from me. (Reading history to understand intentions is a different matter - you should be able to understand a document's behavior just by reading it though.)

This is fairly tolerable for one-off spreadsheets because you're free to throw the entire thing away, import your data again, and have a clean start. A Google sheet which is a living document, that's just a liability in my mind. I know people do this with Excel, and I can only assume it's leagues better than Sheets.

A lot of this would be fixed by using a CSV and/or by using a reduced spreadsheet program that doesn't carry do much baggage & isn't fit for analysis purposes. But at that point I kinda wonder whether SQLite is what you're looking for. You'd be able to do all the functions you'd like, but you could introspect to your hearts content, and there's a huge variety of good tooling to choose from; you wouldn't be forcing everyone to use the same interface. You'd also eliminate a dependency on Google Sheets, which as someone pointed out elsewhere in the thread, might be correlated with your downtime if you're using GKE.

This is super cool, but I'm bummed that it's called "Xls"kubectl and it's for Google Sheets. Something similar should be possible in modern Excel, there's good support for HTTP.

This is so much better than the standard ops tooling. I don’t know why you would ridicule it.

Wow! Based on the title I was prepared to lump this in with the "kill pods via doom" category but this is so much more and honestly inspiring to build on more fun ideas. Congrats on shipping and the content quality of the intro at https://learnk8s.io/real-time-dashboard is fantastic.

Sheetops, I love it.

Oh shitoops

Replacing YAML with spreadsheets has always been our mission as a company, and we will continue to do so.

I mean ... it's not THAT daft an idea tbh

Isn't a spreadsheet just CSV or similar?

For data storage, yes. But most spreadsheet programs (Google sheets, excel, numbers) also allow you to have dynamic data through the use of formulas, using built in functions or straight code of some variety.

The extra functionality is why you hear horror stories of entire multi-million dollar companies runnning their entire logistics pipeline through a single Excel “spreadsheet” and the accompanying brown pants moment of deleting the master doc and not the working (altered) copy.

Also unknown (they can be known, but frequently are not by users) bounds, xls has around a 65k row limit. These are the weird edges where people get burned and a simple csv/jsonl/yaml file is more resilient to scaling/time.

Catchy name

remember the good old days when we used to have words with vouls?

I’m pretty sure it’s spelled “vwls”

On the bright side gsheets won't immediately hit the 65k row limit in og xls files. Eventually someone is going to export or import to xls and truncate a ton of their infra.

I like it. Great and fun! I wish there was more functionality. Like allowing me to list the nodes of the cluster with extra colums like some labels, specific taints, ... Same with pods

I fear someone will take this project seriously.

Spreadsheets are eating the world! :-)

(With apologies to Marc Andreessen)

Reply on news.ycombinator.com