I’ve always thought this would be helpful for small in-house ops scripts and things like that, but unfortunately the logo and terminology makes this a bit to risqué for a corporate environment. I suppose that’s part of the GNU spirit!
+1 for openfaas. I haven’t used it in production, but the option to run an entirely custom container underneath the function opens up the possibility of running a “proper” application underneath (the use case I’ve been meaning to test is a fully-featured rails app)
I think this is a great example of the existential dread I feel when thinking about modern cloud “solutions” — at what point are there too many moving parts? I say this as someone who administers kubernetes clusters in AWS (kops, not EKS). I’m starting to wonder how much of this abstraction and “scale” is needed for 99% of use cases
The caveat with a third party oauth solution is that you are now dependent and reliant on the third party to _let_ you use them to log in. Here are some fun experiences I’ve had with Facebook over the last couple of years:
- Our app was _deleted_ without any notice and any means of appealing (didn’t appear in the appeals page, and of course there’s no human support). We even filed a ticket and were told that they couldn’t help us because the app was “gone” in their system. Luckily we require an email address or we would have completely lost the ability to authenticate a subset of our users.
- A different internal app was banned from using “Facebook Login” because we were “providing a broken user experience” — the app was not even exposed for login in our system. We couldn’t appeal because the warning notice didn’t allow responding from our mailing list. Changing the primary contact didn’t work either, and we even disabled the login on the app just in case. Still revoked with no means of getting it back.
Google has been less awful to work with, but they make you jump through lots of hoops to get public login permissions. In summary, think very carefully about a third party Oauth solution.
Is there a CPU limit or timeout for queries? I’d be a little concerned that an intentionally slow and inefficient query could pin the CPU at 100% and ruin the performance for other users
I think 5 is probably the biggest — in terms of emulation, you can run TK4 in Hercules, but it’s literal decades behind the modern mainframe. How am I supposed to learn if the concepts are hidden behind a massive monetary barrier? Reading manuals can only get you so far
Currently the cluster autoscaler supports using a pool of spot instances based on pricing, which is super helpful for test clusters, and there are some other tools available to ensure that you can evict your spot nodes when amazon needs them back