You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
First thanks to you guys in codecov, providing such an amazing product.
There are a few tips I found may save your time
1. Get a let's encrypt certificate or commercial certificate
At first, I tried to set up an HTTP site for codecov, but what I experienced is login would fail.(maybe I'm wrong about that)
So an HTTPS site is a must, and codecov uploader does not support self-signed certificate, so a let's encrypt certificate would help you.
2. Make sure to add sslmode=require when using timescaledb or pg(if ssl is enabled)
Codecov require a timescaledb to have full feature, if your timescaledb instance is provided by SaaS or a helm chart, there's high probability that it provides service with SSL enabled, if sslmode=require is not added, you would not start codecov properly.
3. Inject RUN_ENV=ENTERPRISE in worker container
If not set, links in pr comment would use app.codecov.io instead of your instance.
See #14 for details
Notice the duplicated reach, I personally advise you to use reach,diff,flags,tree, or there would be two coverage Grid pictures, which would take a lot of space in your pull request.
# for GHE_URL, setting env would be sufficient
ENV REACT_APP_ENV=enterprise
ENV REACT_APP_GHE_URL ="https://your-ghe.example.com"
6. Try the cloud?
I tried to set up codecov instance and make it usable in a week, but I'm not confident to say that this is in perfect condition, maintaining a service could be tiring, codecov contains 4 components and 4 dependencies, they all require basic monitoring, or backup if they are database. And once it went wrong, all the thing you can do is to check the code, or preying that there's someone nice and know this well in the community.
I'm not an employee neither sentry nor codecov, but I do believe for some/most of the people, cloud service would be cheaper and easier.
The text was updated successfully, but these errors were encountered:
Regarding the SSL part. When trying with SSL, I managed to get auth (GitHub) work but could not get it working without. While app settings are identical in both cases. So I'm also tending to believe that SSL is a must-have to make auth work.
First thanks to you guys in codecov, providing such an amazing product.
There are a few tips I found may save your time
1. Get a let's encrypt certificate or commercial certificate
At first, I tried to set up an HTTP site for codecov, but what I experienced is login would fail.(maybe I'm wrong about that)
So an HTTPS site is a must, and codecov uploader does not support self-signed certificate, so a let's encrypt certificate would help you.
2. Make sure to add sslmode=require when using timescaledb or pg(if ssl is enabled)
Codecov require a timescaledb to have full feature, if your timescaledb instance is provided by SaaS or a helm chart, there's high probability that it provides service with SSL enabled, if
sslmode=require
is not added, you would not start codecov properly.3. Inject RUN_ENV=ENTERPRISE in worker container
If not set, links in pr comment would use app.codecov.io instead of your instance.
See #14 for details
4. Override default comment layout
The default pr comment layout would be :
https://github.com/codecov/shared/blob/5e3ba08c6e526846339d467dd8984c40281733c0/shared/config/__init__.py#L46C1-L50
Notice the duplicated
reach
, I personally advise you to usereach,diff,flags,tree
, or there would be two coverage Grid pictures, which would take a lot of space in your pull request.5. Build your own frontend/gazebo
There are plenty of configs that can only be configured during Build time, such as GHE_URL -> https://github.com/codecov/gazebo/blob/5e7147306dc0b297603fe97f2d9fece63d9dd8fa/src/shared/utils/provider.js#L85C1-L87
So if you want to have a normal frontend that can link you back to pull request, the only way to achieve that is build your own front-end.
you can refer to the official dockerfile when building your own image
6. Try the cloud?
I tried to set up codecov instance and make it usable in a week, but I'm not confident to say that this is in perfect condition, maintaining a service could be tiring, codecov contains 4 components and 4 dependencies, they all require basic monitoring, or backup if they are database. And once it went wrong, all the thing you can do is to check the code, or preying that there's someone nice and know this well in the community.
I'm not an employee neither sentry nor codecov, but I do believe for some/most of the people, cloud service would be cheaper and easier.
The text was updated successfully, but these errors were encountered: