Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
xzzy
Mar 5, 2009

That's where we were three years ago, and went down the road of running our own cluster (we chose okd). It works, but it is a lot of work.. especially if you start opening it up to users.

I haven't yet regretted putting our internal apps into it, everything's been solid.

Adbot
ADBOT LOVES YOU

Methanar
Sep 26, 2013

by the sex ghost
Kubernetes is the only viable choice imo.

Its easier today than it used to be by a significant margin with kubeadm being GA for bootstrapping your cluster. It's still a fair amount of work and you need to still do a lot yourself, you can't easily defer the loadbalancing problem to aws but instead need to build your own contraption.

I wouldn't build a full kubernetes deal just for one app though, there are easier ways of getting what you want regarding load balancing: you'd need to solve the load balancing problem in addition to all the kubernetes problems.

freeasinbeer
Mar 26, 2015

by Fluffdaddy
Two performance tips I have for nginx ingress; First is to make sure the service load balancer that fronts the nginx pods is set to External traffic policy local and then that you the load balancer policy is set to ewma.

This is a contrived example; but I have 3 handling 4K-10k rps between them, using half a cpu each. Receiving 40 MBs.

freeasinbeer fucked around with this message at 05:19 on Oct 17, 2020

Methanar
Sep 26, 2013

by the sex ghost

freeasinbeer posted:

Two performance tips I have for nginx ingress; First is to make sure the service load balancer that fronts the nginx pods is set to External traffic policy local and then that you the load balancer policy is set to ewma.

This is a contrived example; but I have 3 handling 4K-10k rps between them, using half a cpu each. Receiving 40 MBs.

These are good suggestions. I'll try them out next week. In my case I'm going to need to be able to handle upwards of 60k to 100k rps, probably worth 4-7 gbps. I'm still leaning now towards trying out the alb ingress controller since there has been so many people name dropping it.

https://kubernetes-sigs.github.io/aws-alb-ingress-controller/guide/controller/how-it-works/

Looking at the architecture diagram alone its delightfully simple and eliminates a whole userland packet processing, and an entire node hop if this controller is intelligent to only have nodeports registered of nodes that actually contain a particular pod rather than yoloing it out and letting iptables deal with it if necessary.




Your post is a bit hard to read, but for everybody else's reading pleasure, this is what is being referred to for tuning nginx

https://kubernetes.io/docs/concepts/services-networking/service/#aws-nlb-support

quote:

Unlike Classic Elastic Load Balancers, Network Load Balancers (NLBs) forward the client's IP address through to the node. If a Service's .spec.externalTrafficPolicy is set to Cluster, the client's IP address is not propagated to the end Pods.

By setting .spec.externalTrafficPolicy to Local, the client IP addresses is propagated to the end Pods, but this could result in uneven distribution of traffic. Nodes without any Pods for a particular LoadBalancer Service will fail the NLB Target Group's health check on the auto-assigned .spec.healthCheckNodePort and not receive any traffic.

https://github.com/kubernetes/ingre...md#load-balance

quote:

Sets the algorithm to use for load balancing. The value can either be:

round_robin: to use the default round robin loadbalancer
ewma: to use the Peak EWMA method for routing (implementation)

Methanar fucked around with this message at 06:48 on Oct 17, 2020

Hadlock
Nov 9, 2004

Hed posted:

Would anyone go down the path of kubernetes for an application hosted on their own on-prem hardware?

Nuke your etcd cluster and get back to me

On prem k8s is possible, but it's not fun

Hughlander
May 11, 2005

Hadlock posted:

Nuke your etcd cluster and get back to me

On prem k8s is possible, but it's not fun

At least if you're going to do it, do it with a private cloud setup. Company I worked for had a few tens of thousands of servers in DCs around the world that were either being migrated or their foot print was being migrated to private cloud to then have groups run k8s on top of.

Gyshall
Feb 24, 2009

Had a couple of drinks.
Saw a couple of things.
I interviewed at a place a few months ago that wanted a single engineer to run the k8s on prem cluster running on hyper v in a managed colo somewhere. Most of the interview was spent just trying to understand why.

Vulture Culture
Jul 14, 2003

I was never enjoying it. I only eat it for the nutrients.
If you're not running Kubernetes at scale, and you have a basic understanding of distributed systems/schedulers and PKI, the operational overhead of K8s is totally overstated. Sure, trying to troubleshoot a non-functioning etcd is a pain in the rear end, but you can also just throw out your cluster and restore the whole thing from Velero in ten minutes.

I'd still prefer a cloud-managed cluster instance any day, but if I was in some airgapped environment that had to be on-prem because of somebody's reasons, I don't think there's reason to be scared of it or choose a worse approach for managing your applications.

Mr. Crow
May 22, 2008

Snap City mayor for life
So Bitbucket is shuttering their self hosted option for some reason and we need to migrate to a new self hosted source control system. Gitlab is the obvious choice but I was curious is anyone uses anything else to success? Just want to weight our options

New Yorp New Yorp
Jul 18, 2003

Only in Kenya.
Pillbug

Mr. Crow posted:

So Bitbucket is shuttering their self hosted option for some reason and we need to migrate to a new self hosted source control system. Gitlab is the obvious choice but I was curious is anyone uses anything else to success? Just want to weight our options

Azure DevOps Server if you don't mind Windows.

Methanar
Sep 26, 2013

by the sex ghost

Mr. Crow posted:

So Bitbucket is shuttering their self hosted option for some reason and we need to migrate to a new self hosted source control system. Gitlab is the obvious choice but I was curious is anyone uses anything else to success? Just want to weight our options

My org is probably moving to gitlab.

Killing selfhosted bitbucket is stupid.

xzzy
Mar 5, 2009

I can't offer any gitlab alternatives but I can say we've been using gitlab for years and I think it's great.

Docjowles
Apr 9, 2009

Yeah Atlassian is forcing everyone to move to the cloud version of their poo poo unless you are willing to pay $Texas for an indulgence. It’s pretty cool :jerkbag:

We run GitLab for about 1000 users and are happy with it. Unless it’s for one small team I recommend springing for one of the paid tiers. The free edition is pretty crippled, missing poo poo like mandatory approvals before merging a pull request and “a working search feature”. You need at least the starter tier to get back near feature parity with Bitbucket.

NihilCredo
Jun 6, 2011

iram omni possibili modo preme:
plus una illa te diffamabit, quam multæ virtutes commendabunt

Mr. Crow posted:

So Bitbucket is shuttering their self hosted option for some reason and we need to migrate to a new self hosted source control system. Gitlab is the obvious choice but I was curious is anyone uses anything else to success? Just want to weight our options

We're using Gitlab and are quite happy with it as well.

Sometimes it's got minor bugs that have been left open for 2+ years in favor of adding more enterprise paid features, which I can't really blame them for. None of those have been show-stoppers, just stuff like the build cache not triggering and slowing down builds by a few minutes.

I might consider Gitea if I did not need a built-in CI/CD system or built-in package manager, and/or if I didn't have a beefy server to host it on. Gitlab is a massive resource hog, while Gitea runs on a Pi and feels blazing fast at all times.

Then again, Gitlab is an enterprise product with all that it entails, e.g. I've never had a single issue running a plain `gitlab backup create && apt-get upgrade` after a new release; whereas Gitea is an open-source project that isn't even dogfooding itself yet (is code hosted on Github).

edit: Gitea apparently supports git mirroring (while it's a paid feature in Gitlab) so you can maybe install both with mirrored repos and get a feel for which one you like better.

Soricidus
Oct 21, 2010
freedom-hating statist shill

Docjowles posted:

Yeah Atlassian is forcing everyone to move to the cloud version of their poo poo unless you are willing to pay $Texas for an indulgence. It’s pretty cool :jerkbag:

switching to the atlassian cloud sounds like a great option. how’s its performance across an airgap?

Hughlander
May 11, 2005

Soricidus posted:

switching to the atlassian cloud sounds like a great option. how’s its performance across an airgap?

It's getting better all the time https://arxiv.org/abs/2004.06195

Hadlock
Nov 9, 2004

We used gitea/gogs for a couple months while some of our senior legacy Java developers were stonewalling our move from SVN to Git, it worked very well for the six or so months we used it

FISHMANPET
Mar 3, 2007

Sweet 'N Sour
Can't
Melt
Steel Beams
GitHub as an on-prem offering as well, GitHub Enterprise Server. The latest version even supports GitHub Actions, so you've got some CI/CD builtin (though we haven't upgraded to that version yet so I don't have any experience with it yet).

The Fool
Oct 16, 2003


I need to add some devopsy/cloud/automation/config management voices to my Twitter feed anyone have some good follow recommendations?

12 rats tied together
Sep 7, 2006

He's not specifically in this space but I recommend @sbellware as a pro twitter follow for anyone working SaaS.

Blinkz0rz
May 27, 2001

MY CONTEMPT FOR MY OWN EMPLOYEES IS ONLY MATCHED BY MY LOVE FOR TOM BRADY'S SWEATY MAGA BALLS
Not a comprehensive list but I follow Seth Vargo (@sethvargo), Charity Majors (@mipsytipsy), Corey Quinn (@QuinnyPig), Mitchell Hashimoto (@mitchellh), and @SimpsonsOps and find them to be pretty good

Matt Zerella
Oct 7, 2002

Norris'es are back baby. It's good again. Awoouu (fox Howl)
Also IanColdwater

freeasinbeer
Mar 26, 2015

by Fluffdaddy
Also the dude behind envoy, who’s name escapes me.

in a well actually
Jan 26, 2011

dude, you gotta end it on the rhyme

Kelsey Hightower, Liz Fong Jones, Erowid Recruiter

whats for dinner
Sep 25, 2006

IT TURN OUT METAL FOR DINNER!

Alex Hidalgo and Amy Tobey are really good too

Blinkz0rz
May 27, 2001

MY CONTEMPT FOR MY OWN EMPLOYEES IS ONLY MATCHED BY MY LOVE FOR TOM BRADY'S SWEATY MAGA BALLS

PCjr sidecar posted:

Kelsey Hightower, Liz Fong Jones, Erowid Recruiter

Oooh yeah forgot about Liz

whats for dinner
Sep 25, 2006

IT TURN OUT METAL FOR DINNER!

Blinkz0rz posted:

Oooh yeah forgot about Liz

Yeah Liz is a real pro follow. She's got some great threads covering off on changes she's made like migrating to Graviton instances and a pretty substantial Kafka uplift.

FamDav
Mar 29, 2008

freeasinbeer posted:

Also the dude behind envoy, who’s name escapes me.

matt klein

The Fool
Oct 16, 2003


12 rats tied together posted:

He's not specifically in this space but I recommend @sbellware as a pro twitter follow for anyone working SaaS.

Blinkz0rz posted:

Not a comprehensive list but I follow Seth Vargo (@sethvargo), Charity Majors (@mipsytipsy), Corey Quinn (@QuinnyPig), Mitchell Hashimoto (@mitchellh), and @SimpsonsOps and find them to be pretty good

Matt Zerella posted:

Also IanColdwater

PCjr sidecar posted:

Kelsey Hightower, Liz Fong Jones, Erowid Recruiter

whats for dinner posted:

Alex Hidalgo and Amy Tobey are really good too

freeasinbeer posted:

Also the dude behind envoy, who’s name escapes me.

FamDav posted:

matt klein

Thanks for all the suggestions guys
Couple of them I'm already following, like Quinn and Coldwater. Adding the rest

SurgicalOntologist
Jun 17, 2004

Anyone familiar with kaniko here? I'm having a poo poo time trying to figure out why nothing is caching. But I'll spare the gory details if no one uses it.

Docjowles
Apr 9, 2009

come to aphyr’s feed for the tech content, stay for the extremely graphic male bdsm pics :v:

Doc Hawkins
Jun 15, 2010

Dashing? But I'm not even moving!


tbh it's surprising sado-masochism isn't more widespread among distributed systems peeps

Zorak of Michigan
Jun 10, 2006


Why take work home with you?

Matt Zerella
Oct 7, 2002

Norris'es are back baby. It's good again. Awoouu (fox Howl)

Zorak of Michigan posted:

Why take work home with you?

Methanar
Sep 26, 2013

by the sex ghost

Zorak of Michigan posted:

Why take work home with you?

working from home is living at work

Gyshall
Feb 24, 2009

Had a couple of drinks.
Saw a couple of things.
I actually write YAML templates in a full gimp suit

whats for dinner
Sep 25, 2006

IT TURN OUT METAL FOR DINNER!

Zorak of Michigan posted:

Why take work home with you?

I figure it's more of a case of turning your hobby into a vocation

Doc Hawkins
Jun 15, 2010

Dashing? But I'm not even moving!


you know what they say, wear a gimp suit to your job and you'll never work a day in your life

NihilCredo
Jun 6, 2011

iram omni possibili modo preme:
plus una illa te diffamabit, quam multæ virtutes commendabunt

I want to create a small webapp for friends & family to use during meetups. Most of the time it will just serve some static content, but once in a while it will need to fire up an external process (a docker container) that could really benefit from some compute oomph.

This is more of a hobby fun project to learn a different webdev stack than something we actually need, so I'd like to run it for free or for literal peanuts.

It looks like I should be able host it all on the Google Cloud free tier. Dockerized webapp in the App Engine (28 hours/day), then upon demand run the compute jobs using Cloud Run (180k vCPU seconds / month), store generated data in Cloud Storage (5GB). I've looked at AWS, Azure, DO, and Heroku and their (permanent) free tiers don't seem to compare. Is there any pitfall I should be aware of?

Adbot
ADBOT LOVES YOU

minato
Jun 7, 2004

cutty cain't hang, say 7-up.
Taco Defender
If it's mostly static content then just throw it in a storage bucket to avoid the web server running at all, and use Cloud Run to initiate the occasional compute.

Only pitfall I can think of is to make sure you use a robots.txt to ensure your static site isn't crawled as it will consume bandwidth you don't need. And maybe set up a Budget Alert so you're aware if you start to approach the free tier threshold.

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply