#random (2020-03)
Non-work banter and water cooler conversation
A place for non-work-related flimflam, faffing, hodge-podge or jibber-jabber you’d prefer to keep out of more focused work-related channels.
Archive: https://archive.sweetops.com/random/
2020-03-03
2020-03-04
@Erik Osterman (Cloud Posse) SCALE already added a comment about the state of emergency declaration:
COVID-19 and SCALE 18x
Submitted by Phil Dibowitz on Thu, 2020-02-13 21:42
UPDATE (March 4, 2020): We wanted to provide an update given there are now six known cases of Coronavirus in the Los Angeles area. While the LA County Board of Supervisors have declared a state of emergency, the goal of that move is to acquire additional funding for prevention and planning. The Mayor of Los Angeles’ office gave a statement this morning and said there is no proof of communal transmissions and importantly, that there are no plans to stop public gatherings. As always we are staying in touch with local officials and will keep on top of all developments and follow all authority guidelines to keep everyone safe. We look forward to seeing you all tomorrow!
Standard advice from virologists still applies: Wash your hands they way you’re supposed to and keep your hands away from your face and don’t stand next to someone who looks like they’re fighting the flu.
in case anyone is petrified… fatality rates are low and extremely low if you’re young.
well… i was super wrong about this
been in quarantine for 3 weeks now
150 total cases in the US, so far 11 fatal
altho the below link is a lil slow compared to above, but gives good insight for real time updates
I’m mostly concerned about my kid that’s just a couple of months old. I’m WFH for the most part due to the Flu going around
Ya, this is my concern as well…
2020-03-06
This looks pretty sweet, pretty certain helm charting this would be not so hard.
dashing in golang
jajaja que bien!
2020-03-07
2020-03-08
2020-03-09
datadog vs newrelic. your thoughs? I’m using newrelic now but datadog seems cheaper in my setup so I’m looking for additional pros and cons of the two?
Not certain on new relic anymore but the last time I used it there was more of a slant towards application monitoring instead of actual infrastructure monitoring.
been a while though
I’ve been implementing datadog via terraform recently and that has proven to be quite useful for lining up the deployed infra with baseline monitors for metrics and such
the integration for datadog is pretty expansive as well
pricing seems to be per host as well, they get a bit dismayed if you just use say azure integration and filter out all of your VMs….
their logging costs aren’t too bad either I suppose
Datadog have an APM product but our experience with it hasn’t been great.
thanks. need to check the datadog apm whether it fits our needs then
How do y’all manage brew packages that you want everyone to have?
I use a Brewfile too. Here’s my personal one with my tf tools.
https://github.com/nitrocode/dotfiles/blob/master/macos/Brewfile#L196
sweet! never knew of brewfile. i feel like i should though based on my use of Taskfile
2020-03-10
2020-03-11
2020-03-12
Don’t get me even started on Cognito .
I have the feeling that everything on aws that is not S3, EC2, Route53 is crappish. Some services more than the others and then cognito :-D
I mean basic functionality works. you can create users and authenticate them. ;D fun starts when you try to e.g. add federated identities or extend attribute schema with some custom attribs which then are not being returned within auth resp
Do you guys prefer Auth0 or something else?
I use keycloak a lot
@Joe Niland hell no. Why do you want to give one of your assets (customer information) to some 3rd party?
we were looking into using https://www.openiam.com/ but it’s not fitting out usecase so the idea to host it ourselves was dropped and we build sso solution based on Cognito serving as an identity provider and oauth2
quick question: does archive.ubuntu.com work for anyone of you?
was able to browse it just now in the browser
and apt updates from [us-east-1.ec2.archive.ubuntu.com](http://us-east-1.ec2.archive.ubuntu.com) work just fine
Wfm
Anyone use https://github.com/pusher/oauth2_proxy? I’ve been using it successfully in k8s with nginx-ingress for several months, and after deploying changes to an app (without touching the oauth2_proxy or ingress settings) I’m getting wierd redirect behavior I can’t figure out.
We used to use it before bitly’s version became EOL
Fork happened later (pusher repo)
most recently @btai has been looking into it
We moved to Keycloak + Gatekeeper instead. More robust (and complicated) solution.
I’ve been quite happy with it, and now I’m just confused.
Confused in what way?
Everything was working perfectly, and without touching it I’m now getting wierd redirects.
Specifically if I go to https://example.com/page I’m being redirected to https://example.com/example.com/page
I made some minor changes to the app running behind the proxy, did not touch the ingress definition or oauth2_proxy config at all.
I’m annoyed that it took me so long to realize, but relieved it was my own fault.
The alternative is… terrifying.
I’d sooo rather find out something is my fault than have something randomly start working again
Exactly! One is just “oh, I’m being a little dumb” the other is “what happened to reality?!”
Way easier to explain to others as well. Telling someone that cloud gremlins are to blame just sounds like you are hiding something….
#anti-cloudgremlins
haha! Yeah, gotta keep that one in reserve for when the cloud gremlins really show up.
2020-03-13
2020-03-16
2020-03-17
2020-03-19
Your project has 30 repos. The origin of the initial commit for 20+ of them are from you….do you take pride or shame in that fact?
Definitely! Unless your job position title is: Director of Creating New Repositories
More context needed. I worked at a company that had over 150 repos. Most of them Java with under 20 commits, the rest DB related. They could definitely be consolidated into ~30 Java repos with 10 being parent repos. I’d give you $100 to slap the person that created all of them, followed by a good public shaming.
the coronavirus rhapsody https://twitter.com/paulomatheny/status/1240477469976731648
2020-03-20
I have 40+ repos, but I try to keep them all single purpose focused to avoid mono repo format, esp with terraform modules. I just recently created a terraform CICD plan that takes a yaml file and creates the repos, establishes branch policies, notifications, merge/squash/rebase options etc. I don’t think count of repos should be too concerning if you have a way to manage them effectively.
Is this also so that to create new repos someone opens a PR and updates the repo YAML config?
Right now I’m trying to convince people to use it so it’s just for my team. As such I’m not doing any pull requests or more strict process. Adoption for any more structure is really just tough right now. I don’t think any of the teams use git with best practices at all right now. I was looking through some history and see no evidence of squashing commits to keep a nice clean history it’s all very messy
I think reducing the overhead to submitting something is really important right now if I want to gain even this small win
The main reason I like to create separate repos is for simplifying potential Azure DevOps pipelines. I might reconsider this soon and try to do more of a mono repo format for anything plan related, but most of the time it’s easier just to isolate the work into small repos for specific purposes
makes sense
2020-03-21
@sheldonh what did you use to parse your yaml file for such a construct? Willing to share?
Sure! Basically what I did is I set up a QA and a prod folder. Each of the plans for each environment resided in this. This is different than some examples where you see a plan with a QA and prod folder in its own location.
Then in the parent QA folder I had a yellow file for default settings, and an override yaml in the folder below. Since merge function overrides existing matches you can override any default value in the child folder if required, otherwise you don’t have to define anything extra
I use the yamldecode function and merge that are built into terraform
After having done this more complicated project, I’m mixed on whether or not I would do it this way again. If I modify this I might consider just setting these as locals and having the overrides also be a local file, or more likely leverage auto.tfvars . Terraform cloud is a bit confusing so I didn’t know about the auto overrides option at the time.
If I was doing this again I’d say explore QA.auto.tfvars. Would be a more standard way and what I’d likely do in the future
2020-03-22
We are doing something similar with our pipeline. We have developed a pretty nice set of ansible modules for interacting with the github api, and are hoping to open source these. The yaml definitions also control deployment of artifactory permissions (and repositories, eventually), and credential injection into SSM for CI.
2020-03-26
does someone here know a similar tool like jq for yaml?
There are a couple different tools called yq - I prefer this one: https://github.com/kislyuk/yq
It does require jq, as it’s just a jq wrapper
I saw that one as well as oq which seems to have less features and is a rewrite.
Thanks !
I’ve also seen https://github.com/mikefarah/yq - might be worth a look as it has some sed-like commands that could be useful
thanks that seems even better for my use case
Yep, we use the one by mikefarah and distribute it here with our alpine packages https://github.com/cloudposse/packages/tree/master/vendor/yq
2020-03-27
Adding @discourse_forum bot
@discourse_forum has joined the channel
2020-03-29
@Erik Osterman (Cloud Posse) you’re using the slack-export-viewer for the archive.cloudposse.com no? I saw you had some issues with the original repo, I tried to use the cloudposse one but I keep having issues with the “TypeError: string indices must be integers” any hint?
No we wrote our own
Couldn’t get that one to work and it was over-architected
Well good to know! Thanks for the info
2020-03-30
This looks really interesting as a self hosted alternative to Zapier and similar https://www.furnace.io/
any suggestions I need to build a microservice infrastructure. Need to have the following.
- Postgres Database
- Redis Database
- Web Application
Create a zero downtime auto deployment plan for the web application and microservice
2020-03-31
how ‘zero downtime’? ^^