PyAr / pyar_infra

Python Argentina Infrastructure

Python Argentina Infrastructure

This is the repository with all the code and documentation to handle PyAr infrastructure

We are working with kubernetes in Azure aks

We use Helm as a package manager.

Step-by-step guide to deploy de cluster

HTTPS config

We are using HTTPS with Let's Encrypt

Settings details at: https://docs.microsoft.com/en-us/azure/aks/ingress

Redirecter.

We have lot of domains. But python.org.ar is our principal.

To handle redirects from other domains we have two models:

1. nginx ingress rules, different services configured with files in the redirects directory

   • first time:

     kubectl create -f redirects/prueba.yaml

   • after any change:

     kubectl apply -f redirects/prueba.yaml

   • to see what's there:

     kubectl get pods --namespace=ingress-basic

2. nginx server, handling redirecter.python.org.ar, the configuration is stored in a config-map: stable/pyar-rewrites/templates/config_map.yaml

   • to deploy it run:

helm upgrade --install --wait pyar-rewrites stable/pyar-rewrites

- check its public IP (`LoadBalancer Ingress` after doing `kubectl describe services pyar-rewrites`) and load it in Cloudflare for the `redirecter` entry.

The Database, a PostgreSQL cluster

Using https://github.com/helm/charts/tree/master/stable/postgresql

Deploy:

El siguiente comando hace el deploy. NOTA: NO tiene que estar el secreto pgcluster-postgresql al momento de deployar PSQL (se crea en ese proceso).

helm upgrade --install --wait -f values/production/postgres_cluster.yaml pgcluster oci://registry-1.docker.io/bitnamicharts/postgresql

This cluster is using a PersistentVolumeClaim and a "lock" is created manually in azure to prevent unintencional deletes. Detail about locks: https://docs.microsoft.com/en-us/azure/azure-resource-manager/resource-group-lock-resources

To set the backup:

kubectl apply -f k8s/pgsql_bkps_jobs/pg-storage-class.yaml
kubectl apply -f k8s/pgsql_bkps_jobs/pg-persistent-volume-claim.yaml
kubectl apply -f k8s/pgsql_bkps_jobs/pg-backup-cronJob.yaml

Connect to the cluster

# get the password
export POSTGRES_PASSWORD=$(kubectl get secret --namespace default pgcluster-postgresql -o jsonpath="{.data.postgres-password}" | base64 --decode)
# connect
kubectl run pgcluster-postgresql-client --rm --tty -i --restart='Never' --namespace default --image docker.io/bitnami/postgresql:11.5.0-debian-9-r84 --env="PGPASSWORD=$POSTGRES_PASSWORD" --command -- psql --host pgcluster-postgresql -U postgres -p 5432

Configuration

We have to create the databases and users manually

Restore backups

1. Download the Backup file from Azure Blob Storage
2. Create a console to the PostgreSQL cluster

# get the password
export POSTGRES_PASSWORD=$(kubectl get secret --namespace default pgcluster-postgresql -o jsonpath="{.data.postgres-password}" | base64 --decode)
# connect
kubectl run pgcluster-postgresql-client --rm --tty -i --restart='Never' --namespace default --image docker.io/bitnami/postgresql:11.5.0-debian-9-r84 --env="PGPASSWORD=$POSTGRES_PASSWORD" --command -- /bin/bash

1. On a new local console, copy the local downloaded file to the cluster

kubectl cp *.dump pgcluster-postgresql-client:/tmp/backup

1. On the existing console to the PostgreSQL cluster run the restore command. Change the CHANGE_THE_DATABASE for the correct value

I have no name!@pgcluster-postgresql-client:/$ pg_restore --host pgcluster-postgresql -U postgres --d CHANGE_THE_DATABASE /tmp/backup

Python Argentina community website

http://www.python.org.ar

helm upgrade --install --wait --timeout 120s --values values/production/pyarweb.yaml pyarweb-production stable/pyarweb

Wiki

Using https://github.com/helm/charts/tree/master/testing/wiki

Staging:

helm upgrade --install --wait --timeout 120s --values values/staging/pyar-wiki.yaml staging-wiki stable/pyar-wiki --debug

Production:

helm upgrade --install --wait --timeout 120s --values values/production/pyar-wiki.yaml prod-wiki stable/pyar-wiki --debug

Asociación Civil administration. (asoc_members)

Production

helm upgrade --install --wait --timeout 120s --values values/production/asoc_members.yaml production-admin stable/asoc-members

Join Captcha bot

helm upgrade --install  --wait --timeout 120s --values values/production/join_captcha_bot.yaml captcha-bot-production stable/join_captcha_bot

Once up, talk through Telegram with the bot itself and issue: /allowgroup add CHAT_ID (the CHAT_ID can be seen in the logs doing something similar to kubectl logs captcha-bot-production-5d99c5595d-8wcbb).

Events site (EventoL)

https://eventos.python.org.ar

Events site, using EventoL. We use it to host PyDays, PyCon, Pycamp and other events.

Staging

See instructions on how to copy the static files to Azure, but bear in mind that they might collide with production ones.

Final command to deploy to staging:

helm upgrade --install  --wait --timeout 60000 --values values/staging/eventol.yaml staging-eventos stable/eventol

Production

1. We need to get the static files from Eventol docker image and push them to azure. First get a docker container with all the statics inside (the version might change, check the values/production/eventol.yaml file):

$ docker run --name eventol -it registry.gitlab.com/eventol/eventol/releases:v2.3.4 /bin/ash

2. In other terminal, bring those static files to your dev machine:

$ docker ps
CONTAINER ID   IMAGE             COMMAND      CREATED          STATUS              PORTS      NAMES
2e88bd843642   eventol/eventol   "/bin/ash"   41 seconds ago   Up About a minute   8000/tcp   eventol
$ mkdir /tmp/eventolstatic
$ cd /tmp/eventolstatic
$ docker cp CONTAINER_ID:/usr/src/app/eventol/static .

(at this point you can let go the container from (1) )

3. Upload the static files to Azure Storage

cd /tmp/eventolstatic/
az storage copy -s static -d 'https://pyareventol.file.core.windows.net/eventol-prod-static/' --recursive

3. Deploy to Production:

helm upgrade --install  --wait --timeout 120s --values values/production/eventol.yaml production-eventos stable/eventol