Hi folks!
Is anyone here using at least a few different AWS resources, possibly even through Terraform, and has some experience with AWS and also is okay using a small amount of their AWS credits to test something?
This is #mgmtconfig related of course!
If so, please reach out! You may enjoy this. Thanks
It's looking like my idea to run etcd over SSH for #mgmtconfig was kind of prescient:
https://github.com/etcd-io/etcd/commit/ebc2858f36e736a87c54483fdc588e422dc7628a
Signed-off-by: Ivan Valdes <[email protected]>
@jzb I do agree and your last paragraph here is 💯 on the mark.
Sadly I'm not convinced that Debian is a good user experience.
We're losing the quality/features/experience war because a shiny mac book actually has really good battery life and a pretty good user experience.
People unwilling to pay for Free Software is a major unsolved problem. I think I know a lot about this space both with #mgmtconfig and also with working because of working at AWS. It's sad :(
Automation folks often needlessly special-case scenarios instead of engineering better solutions.
Today I replaced the above, verbose, multi-case code which uses the #mgmtconfig "Network UPS Tools" module with the below shorter code which now works on *any* machine!
This code automates deployment of UPS monitoring and now works with whichever UPS happens to already be plugged in!
@zhenech @jpmens @youam @kevin If I'm understanding all of this correctly, you've got an elegant way to reliably boot off an effective software raid1 EFI boot drive.
I'd like to build this engineering into the open source #mgmtconfig provisioning system and am happy to pay you for your time to walk me through it and get it right. Lmk if you'd have a moment to chat!
I've mostly finished building a new #mgmtconfig module to setup and manage "Network UPS Tools". It was really elegant and fun to write and despite the software being quite old and a bit archaic, it does seem to work well!
If you'd like to test this out, drop me a line!
Wow, more time spent on this than I care to admit, but, it turns out you can't use eth* names on Linux anymore if you actually care about automation! TIL.
Default device names for anything #mgmtconfig manages are now going to be meth0, meth1, methN, unless someone has a better idea.
https://github.com/systemd/systemd/issues/16665#issuecomment-669167184
I just found out go-systemd is actually polling :/
If someone wants to fix this, I know at least #mgmtconfig folks would appreciate it!
#mgmtconfig demo by @littleidea from #cfgmgmtcamp is now public:
James Just James