​​2017-10-07 to 2018-02-25 travails/victory log

​​Backlog of quality issues/optimizations to make:

​​​🤢​ Backend: Prod/stage not DRY w/ secret fetching logic

​​noted 2018-02-18
​​

​​​🤢​ Backend: Lots of magic values, like directories and so on.

​​noted 2018-02
​​* magic values in backend, like for lots of dirs: might want to save these values in a bash env file to source; that way I only have to change a path in one place rather than who knows how many places
​​

​​​🤢​ Backend: hard coded db endpoints (more magic values: put in module instead, or use secret)

​​noted 2018-02
​​* replace hard coded values for db  endpoint in django config with secret value; requires updating secrets for swarms (don't  do thisfor stage though - stage secret is relative to manager, not swarm task)
​​

​​​🤢​ CI/CD: playbooks may be safer if they prompt for values

​​noted 2018-02
​​`* playbooks could  be written to ask for plays instead of requiring extra var from cmd line.

​​​🔴​ CI/CD: Current solution for updating secrets requires downtime (take down stack and redeploy)

​​

​​Python multithreading disabled with uwsgi

​​

​​​🤢​ Push.yml for compose,secrets fails at move step/tar step if compose file/secret dir isn’t existing on remote node.

​​TASK [push-secrets : Move current secrets to prev dir] *****************************
​​fatal: [mm-prod-manager1]: FAILED! => {"changed": true, "cmd": "(cd /home/docker/secrets/ && tar c .) | (cd /home/docker/prev-secrets/ && tar xf -)", "delta": "0:00:00.002387", "end": "2018-02-13 18:29:23.274442", "msg": "non-zero return code", "rc": 2, "start": "2018-02-13 18:29:23.272055", "stderr": "/bin/sh: cd: line 1: can't cd to /home/docker/prev-secrets/\n/bin/sh: cd: line 1: can't cd to /home/docker/secrets/", "stderr_lines": ["/bin/sh: cd: line 1: can't cd to /home/docker/prev-secrets/", "/bin/sh: cd: line 1: can't cd to /home/docker/secrets/"], "stdout": "", "stdout_lines": []}
​​        to retry, use: --limit @/home/rona/projects/moshimoji/ansible/push.retry
​​
​​

​​Not sure if I’m using load balancer for stage - have 2 a records for stage, both ipv4when it should be a cname to lb.

​​

​​​🤢​ STAGE set to ’0’ instead of 0 in be task: unideal solution implemented, which is to cast the value into int in wsgi.py check.

​​noticed 2018-02-15
​​

​​​⚠️​ Logins insecure till SSL implemented

​​noted 2018-02-18
​​

​​​⚠️​ Generate fragment runs against stage endpoint of gql all the time, instead of inferring the environment and going from there.

​​

​​This doc has gotten too heavy! I’m moving to a new one.

​​
​​
​​

​​2018-02-23

​​Git updates
  • ​​1b751ae Backup work to add checks to migrate, sync (amend later)
  • ​​
​​
​​Not sure if need to run the migrate with every stage deploy, or only the data sync.
​​

​​TypeError: e.allUserStatuses is undefined => broken browser render even after doing migrate and sync onto stage

​​f[picking up 6:30 pm]
​​