`agentsdk` depends on `agent/proto` because it needs to get the version to dial.
Therefore, the conversion routines need to live in `agentsdk` so that we can convert to and from the Manifest.
I briefly considered refactoring the agent to only reference `proto.Manifest`, but decided against it because we might have multiple protocol versions in the future, its useful to have a protocol-independent data structure.
Fixes#8218
Removes `wsconncache` and related "is legacy?" functions and API calls that were used by it.
The only leftover is that Agents still use the legacy IP, so that back level clients or workspace proxies can dial them correctly.
We should eventually remove this: #11819
This PR updates the Agent API to use the appearance.Fetcher, which is set by entitlement code in Enterprise coderd.
This brings the agentapi into compliance with the Enterprise feature.
The new Agent API needs an interface for ServiceBanners, so this PR creates it and refactors the AGPL and Enterprise code to achieve it.
Before we depended on the fact that the HTTP endpoint was missing to serve an empty ServiceBanner on AGPL deployments, but that won't work with dRPC, so we need a real interface to call.
The original test is bugged in that it
1. creates a new AGPL coderd with a new database, so no appearance is set in the DB.
2. overwrites the agentClient so the assertion after removing the license is against the AGPL coderd
fixes#10533
refactors `codersdk` workspace agent dialer to use a single websocket connection to the tailnet v2 API for both coordination and DERPMap updates, rather than separate websockets (and the v1 API for DERPMaps).
#7439 added global caching of RBAC results.
Calls are cached based on hash(subject, object, action).
We often use dbauthz.AsSystemRestricted to handle "internal" authz calls, and these are often repeated with similar arguments and are likely to get cached.
So a transient error doing an authz check on a system function will be cached for up to a minute.
I'm just starting off with excluding context.Canceled but there's likely a whole suite of different errors we want to also exclude from the global cache.
Ok, so my last attempt at a fix here failed
https://github.com/coder/coder/actions/runs/7666229961/job/20893608286
I have a new theory: it's not the `terraform` binary that's busy, it's actually `fake_cancel.sh` and it gets marked busy when we `exec` it from the script we write.
Use of `exec` also replaces the executing code in place, rather than starting a new process/shell, so that's why the error we get says `terraform` is busy.
Spike Curtis