p_l

40 ms·

931.
by p_l 2y ago
The easy baremetal cluster these days is k3s. Includes working out of the box ingress controller.
932.
by p_l 2y ago
Doing it well IMO requires not deploying everything as sidecar but maybe, maybe, deploying it as shared node service. In fact pretty sure I've read a write up from Alibaba? on huge wins in performance due to moving Istio out of sidecar
933.
by p_l 2y ago
Things get iffier when you start doing things like running multiple instances of something (maybe you're sticking two test environments for your developers), or suddenly you grew a bit or no longer fit on the server and start migrating
934.
by p_l 2y ago
Well, since it's cloud run... ... you spin knative on kubernetes cluster and copy over the actually-k8s-manifests over :P
935.
by p_l 2y ago
1. Do some capacity planning 2. Apply appropriate changes to application resources (like parameters for spreading pods around) 3. Add descheduler[1] or similar tool to force redistribution of pods 4. Configure your cluster autoscaling param
936.
by p_l 2y ago
Cloud Run explicitly uses KNative APIs... which are kubernetes objects.
937.
by p_l 2y ago
It's not as much mentioned in the docs these days, and Google might have built a more optimized version, but it's still using knative APIs and originally was essentially KNative deployed on GKE just managed by Google.
938.
by p_l 2y ago
The script break so hard is why one of my first times i felt a "grey beard moment" (with actual grey beard even) was telling a junior the "running ansible from their laptop is not acceptable solution"
939.
by p_l 2y ago
Doing the same, grabbed a reasonably cheap Ryzen (zen2) server with 64GB ECC and 4x NVMe SSDs (2x 512G + 2x 1024G). Runs pretty much this stack: "Infrastructure": - NixOS with ZFS-on-Linux for as 2 mirrors on the NVMes -
940.
by p_l 2y ago
TL;DR someone didn't learn to plan capacity.
941.
by p_l 2y ago
There was breaking change around 1.18, which was spread over few releases to make migration easier. Similar fix pattern as with graduating beta to stable APIs for things like Ingress, they just IIRC covered all the core APIs or so? Don&#x27
942.
by p_l 2y ago
Thanks to kubernetes "flattening" that mess into somewhat flat object map (I like to call it a blackboard system :D) it can be reasonably easy to figure out what's the desired state and what's the current state for g
943.
by p_l 2y ago
As I hope decent system administrator... turns out I find running minimal kubernetes setup (k3s is great there) more sensible and cheaper than managing the server in classic way.
944.
by p_l 2y ago
kubernetes is how I keep compute costs in 2-3 digits :V
945.
by p_l 2y ago
If you need to care about state with systemd you still have the "nonsense" of persistent volume claims, they are just something you keep in notes somewhere, in my experience usually in heads of the sysadmins or an excel sheet or a
946.
by p_l 2y ago
Tiny startups don't have money to spend on too much PaaS or too many VMs or faff around with custom scripts for all sorts of work. Admittedly, if you don't know k8s, it might be non-starter... but if you some knowledge, k3s plus c
947.
by p_l 2y ago
I currently use k8s to control bunch of servers. The amount of work/cost of using k8s for handling them in comparison to doing it "old style" is probably negative by now.
948.
by p_l 2y ago
K8s was designed around deployment on premise on bare metal hardware. The cloud extensions were always just a convenience.
949.
by p_l 2y ago
You misspelled DCOS /s
950.
by p_l 2y ago
Using a language that is actually good at DSLs means you can "extend" the language, so all the tooling for it still works with your DSL, too. Then your DSL is essentially just library.
951.
by p_l 2y ago
It's not about backups - it's about loss of ongoing use infrastructure
952.
by p_l 2y ago
TL;DR if you make a DSL, do it right. Which probably requires making them in a language that is prone to extensibility to so you can just reuse existing tooling. Instead a lot of DSLs happen the way IBM JCL happened - it was explicitly not
953.
by p_l 2y ago
And SpaceX exists because US government, specifically Department of Defense, had a program to throw lots and lots of money at companies with a goal of having few of them survive as new space launch vendor.
954.
by p_l 2y ago
It got obsoleted because of bugs in behaviour, IIRC
955.
by p_l 2y ago
Lots of stories about having google account connected to adwords locked out for various reasons, not all valid.
956.
by p_l 2y ago
In Poland, cybertruck as currently sold is at the very edge of acceptable mass for B-class driver license (aka the "normal" car driver license, not truck one). However that comes with addendum that you need to have that license
957.
by p_l 2y ago
It's more that it's CSP from early to mid 1990s, over which Go just added syntax sugar
958.
by p_l 2y ago
Having tried on other platforms, it's not Bazwl, it's not even just Google. It's python packaging and the way the only really supported binary distribution method of Tensorflow for many many years was to use Pip and hope it
959.
by p_l 2y ago
Ousterhout also I think had outsized presence in early VLSI tooling, which is why TCL happened and why is it still common in tooling (pun intended)
960.
by p_l 2y ago
TL;DR the enforcement simply lacks manpower, and the most egregious cases go to court which also takes time.

More ›