Conversation
|
So did I read it correctly.. we are using traefik for scaling it up? how does it work exactly and have we used it before? |
|
No, we have not used traefik before. Howerver, I personally have a lot of experience with it, and it I think it provides several improvements over nginx for this use case
In order to build a reproducible deployment, we'd have to either write ansible for nginx, use an nginx reverse proxy container, or use traefik. Alternatives: It works by just doing round-robin load balancing across however many running containers there are. |
BryonLewis
left a comment
There was a problem hiding this comment.
A couple of questions regarding traefik.
docker/docker-compose.yml
Outdated
| image: traefik:v2.4 | ||
| container_name: traefik | ||
| command: > | ||
| --api.insecure=true |
There was a problem hiding this comment.
totally new to traefik so maybe need a clarification here on the purpose of API besides debugging and the dashboard. Is it needed for load balancing.
https://doc.traefik.io/traefik/operations/api/
There was a problem hiding this comment.
No. We can disable it.
docker/docker-compose.yml
Outdated
| container_name: traefik | ||
| command: > | ||
| --api.insecure=true | ||
| --serversTransport.insecureSkipVerify=true |
There was a problem hiding this comment.
if we eventually remove nginx guess we will want this disabled for prod?
There was a problem hiding this comment.
This is a slightly misleading argument. It allows inscure SSL behind the proxy, not in front. So like if girder were sending self-signed certs. It can also be removed
We have a lot of CPUs that we can throw at providing better response times. Let's scale girder up.
This is probably what we should have done instead of provisioning a second server in AWS for the last demo.
Also enables rolling restart during updates for zero downtime a535e44
All the pretty colors