Tuesday, October 8, 2019

  • My 15(?) year old wix website has finally been cancelled (which is probably for the better).
  • "And the day came when the risk to remain tight in a bud was more painful than the risk it took to blossom." - Anais Nin
  • Finished Once Upon a Time in Hollywood.
  • Smoked the pastrami beef ribs that had cured for 6 days. Amazing flavor.
    • Next set of beef ribs, I want to try wrapping earlier, closer to the time you would for brisket. I'm curious how much it changes tenderness vs bark.
  • Paid off the remaining balance on the sequoia hospital bill from years ago; was like $220.
  • Cancelled the amex.
    • They tried to give two incentives, one for miles (20k) and one for cashback ($175), both requiring that you spend 3k in 3 months.
    • The companion pass is forfeited (I had 1), and the annual fee is not prorated (there were 4 months left in that this billed year).
    • Removed my amex from google play. Removed the amex browser bookmark. Mint automatically made it inactive, I didn't have to do anything!
  • I would like the Vitamix A3500 - it's so much better than my ninja, and I use it at least once daily. It's $525. I'll probably wait until after I move.
  • The Strassburg sock is definitely helpful for plantar faciitis. WFH makes it a lot easier to get hours in during the day, since I find it annoying to fall asleep wearing.
  • Put the 2020 reg on the Ducati. It's yellow, so it looks better than the blue 2019 sticker.
  • Watch Rob Zombie's new 3 From Hell, the third installment in the series after House of 1000 Corpses -> The Devil's Rejects.
  • The carolina reaper single chip is sold out online, but I just discovered that the riteaid on aviation/artesia has these 3 flavors:
    • I like the haunted ghost pepper the best, then salsa verde, then chile limon.
  • Kubernetes/GCP.
    • Good article on kubernetes: https://www.digitalocean.com/community/tutorials/an-introduction-to-kubernetes. Very useful for orchestrating many containers across many machines together. Metrics, storage options, scalability, ML resources, much more through the full GCP suite.
    • Created a project (supercontest) on GCP and removed the default API Project and My First Project.
    • Enabled the Google Compute Engine API for this project.
    • n1-standard-1 machine, just 1 vCPU and 3.75G mem (the smallest standard machine).
    • Billing is by resource. It's like 3 cents per cpu hour and 0.5 cents per gb mem hour, so the n1-standard-1 is about 5 cents per hour, or about a dollar a day, or about $30/mo. My digital ocean droplet is $5/mo.
    • With the $300 promotion on signup, looks like I can test this for free pretty easily.
    • Ended up staying with digital ocean for now. I don't need any of the other GCP products (compute, storage, ml, etc). The one attractive feature is scalability, but I can upgrade with DO as well.
  • GPG key = GNU Privacy Guard.
  • Sentry.
    • Created account, added the flask integration to supercontest.
    • For the free plan, you can record 5,000 errors per month and it keeps history 1 month back. That's 160 errors a day.
    • Linked my gitlab project to the sentry api, created auth token.
  • GitLab.
    • If your project is public, and you have gitlab ultimate, then you get a security dashboard and dependency list. Pretty cool, but I obviously can't take advantage of this.
    • In general, I've been very pleased with the experience on this platform. Moreso than the atlassian stack, jenkins, etc.
    • Added the deploy stage to my yml. Added the variable SSH_PRIVATE_KEY (for the droplet) to gitlab. Added it as a known host directly through ssh-keyscan.
      • Gitlab runs the deployment as root, and the deployment has ansible pull new source on southbaysupercontest.com, so I had to make sure the machine had the proper ssh keys under the root account (not just mine). sudo cp -r ~/.ssh/ /root
    • Ansible doesn't maintain docker images officially anymore, but some user groups keep them around (albeit not very current). Also...you have to specifically install ansible after you pull the image lol.
    • Did not integrate prometheus; there's already a metrics agent that digitalocean provides to its droplets. I put a link to the external DO dashboard in gitlab tho.
    • Updated the readme to indicate where metrics were, as well as SCM, CI/CD, etc. The GitLab and DigitalOcean logos (clearbit) look good.
    • Modified my view to be fluid.
    • The minutes you've used in CI/CD are at https://gitlab.com/profile/pipeline_quota.
    • Integrated with sentry so that error tracking was in gitlab as well.
    • I don't need to use the container registry associated with the project, although this feature is great. My workflow is not build -> test -> pushImage -> deploy, so a record of all the image artifacts in a registry is less important. I test the source and then build the image right at deploy time, which is more than satisfactory until I start scaling the test/review app portion of things.
  • Sometimes dpkg will prompt for config changes (like apt does for install y/n). Use -o DPkg::Options::="--force-confnew"
  • Changed my mem usage alert on the droplet from 90 to 95%.