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
We experienced frequent issues with yggdrasil, such that it is not really usable as backbone for the current gateway infrastructure on the zos nodes. To this end, webgw was created to improve the connection between the backend service and the gateway node. We now need to change the implementation of gateway nodes to use the new webgw isntead.
This means that changes are needed in zos (cc @muhamadazmy ). Next to this, we might also want to include the client in images we deploy with the current gateway infrastructure, otherwise people will have to do this themselves. Also, as part of this change, we want to bill gateway traffic.
Required work
General
Add the "hashed secret" needed to configure the webgw in the workload of a name/subomain
Images will need to be updated to include the client
zos
Remove existing traeffik in favor of webgw
Add webgw
Implement configuration of webgw, when a new (sub) domain is reserved, a secret will have to be configured
Add billing for webgw traffic. There are counters in place already (as prometheus metrics) for data going to individual backends (based on domain name) in webgw.
Terraform/weblets
Update reservations to include the hashed secret
Caveats
Known problems in the deployment:
Subdomain TLS termination needs to be done in the client node now
There is no backward compatibility with the current gateway infrastructure
Existing deployments using gateways will break
There is a manual step involved now when using a gateway from the user side (needs to create a config for the client with the gateway address + secret used)
Existing images won't have the client, until they are updated users need to download the client manually, which might lead to problems for inexperienced users
Community won't like network billing, an attempt to bill private network usage was reverted in 3.8
The text was updated successfully, but these errors were encountered:
Let's start by adding estimates on the work required and potential problems if any are forseen, to get a good overview of the scope of the work required.
Intro
We experienced frequent issues with yggdrasil, such that it is not really usable as backbone for the current gateway infrastructure on the zos nodes. To this end,
webgw
was created to improve the connection between the backend service and the gateway node. We now need to change the implementation of gateway nodes to use the newwebgw
isntead.This means that changes are needed in zos (cc @muhamadazmy ). Next to this, we might also want to include the client in images we deploy with the current gateway infrastructure, otherwise people will have to do this themselves. Also, as part of this change, we want to bill gateway traffic.
Required work
webgw
in the workload of aname
/subomain
webgw
webgw
webgw
, when a new (sub) domain is reserved, a secret will have to be configuredwebgw
traffic. There are counters in place already (as prometheus metrics) for data going to individual backends (based on domain name) inwebgw
.Caveats
Known problems in the deployment:
The text was updated successfully, but these errors were encountered: