Alternate fix for https://github.com/coder/coder/issues/18080
Modifies wsbuilder to complete the provisioner job and mark the
workspace as deleted if it is clear that no provisioner will be able to
pick up the delete build.
This has a significant advantage of not deviating too much from the
current semantics of `POST /api/v2/workspacebuilds`.
https://github.com/coder/coder/pull/18460 ends up returning a 204 on
orphan delete due to no build being created.
Downside is that we have to duplicate some responsibilities of
provisionerdserver in wsbuilder.
There is a slight gotcha to this approach though: if you stop a
provisioner and then immediately try to orphan-delete, the job will
still be created because of the provisioner heartbeat interval. However
you can cancel it and try again.
Currently, the prebuilds documentation states:
```
### Managing resource quotas
Prebuilt workspaces can be used in conjunction with [resource quotas](../../users/quotas.md).
Because unclaimed prebuilt workspaces are owned by the `prebuilds` user, you can:
1. Configure quotas for any group that includes this user.
1. Set appropriate limits to balance prebuilt workspace availability with resource constraints.
If a quota is exceeded, the prebuilt workspace will fail provisioning the same way other workspaces do.
```
If you need to have a separate quota for prebuilds as opposed to regular
users, you are required to create a separate group, as quotas are
applied to groups.
Currently it is not possible to create a separate 'prebuilds' group with
only the prebuilds user to add a quota. This PR skips the org membership
check specifically for the prebuilds user when patching a group.
Fixes https://github.com/coder/coder/issues/17840
NOTE: calling this out as a breaking change so that it is highly visible
in the changelog.
* CLI: Modifies `coder update` to stop the workspace if already running.
* UI: Modifies "update" button to always stop the workspace if already
running.
My understanding is that `io.EOF` is eventually expected, so logging it
as an error may be confusing. For other errors we should definitely
WARN.
```
[info] provisionerd-ip-172-31-12-44-14: recv done on Session session_id=22b9ef8a-9cd6-4188-98e0-573a50d724cc error=EOF
```
Relates to https://github.com/coder/coder/issues/17818
Note: due to limitations in `cobra/serpent` I ended up having to use `-`
to signify absence of provisioner tags. This value is not a valid
key-value pair and thus not a valid tag.
Relates to https://github.com/coder/coder/issues/15432
Ensures that no workspace build timings with zero values for started_at or ended_at are inserted into the DB or returned from the API.
Relates to https://github.com/coder/coder/issues/15432
* Adds a storybook entry for zero values in provisioner timings.
* Coerces a 'zero' start time to an 'instant'.
* Improves timing chart handling for large timeframes. Previously, this
would have caused the tab to run out of memory when encountering a
`time.Time{}`.
* Render 'instants' as 'invalid' in timing chart.
Builds on https://github.com/coder/coder/pull/17570
Frontend portion of https://github.com/coder/coder/tree/chat originally
authored by @kylecarbs
Additional changes:
- Addresses linter complaints
- Brings `ChatToolInvocation` argument definitions in line with those
defined in `codersdk/toolsdk`
- Ensures chat-related features are not shown unless
`ExperimentAgenticChat` is enabled.
Co-authored-by: Kyle Carberry <kyle@carberry.com>
* Refactors toolsdk.Tools to remove opaque `map[string]any` argument in
favour of typed args structs.
* Refactors toolsdk.Tools to remove opaque passing of dependencies via
`context.Context` in favour of a tool dependencies struct.
* Adds panic recovery and clean context middleware to all tools.
* Adds `GenericTool` implementation to allow keeping `toolsdk.All` with
uniform type signature while maintaining type information in handlers.
* Adds stricter checks to `patchWorkspaceAgentAppStatus` handler.
* Updates default Coder prompt.
* Skips the directions to report tasks if the pre-requisites are not
available (agent token and app slug).
* Adds the capability to override the default Coder prompt via
`CODER_MCP_CLAUDE_CODER_PROMPT`.
https://github.com/coder/coder/pull/17163 introduced the
`workspace_app_statuses` table. Two of these fields
(`needs_user_attention`, `icon`) turned out to be surplus to
requirements.
- Removes columns `needs_user_attention` and `icon` from
`workspace_app_statuses`
- Marks the corresponding fields of `codersdk.WorkspaceAppStatus` as
deprecated.
Fixes two issues with the MCP server:
- Ensures we have a non-null schema, as the following schema was making
claude-code unhappy:
```
"inputSchema": { "type": "object", "properties": null },
```
- Skip adding the coder_report_task tool if an agent client is not
available. Otherwise the agent may try to report tasks and get confused.
- Refactors existing `mcp` package to use `kylecarbs/aisdk-go` and moves
to `codersdk/toolsdk` package.
- Updates existing MCP server implementation to use `codersdk/toolsdk`
Co-authored-by: Kyle Carberry <kyle@coder.com>
Fixes https://github.com/coder/internal/issues/272
* Increases healthcheck timeout in tests. This seems to be the most
usual cause of test failures.
* Adds a non-nilness check before caching a healthcheck report.
* Modifies the HTTP response code to 503 (was 404) when no healthcheck
report is available. 503 seems to be a better indicator of the server
state in this case, whereas 404 could be misinterpreted as a typo in the
healthcheck URL.
Should hopefully fix https://github.com/coder/internal/issues/282
Instead of picking a random port for the prometheus server, listen on
`:0` and read the port from the CLI stdout.
Fixes https://github.com/coder/internal/issues/329
This was due to a race between the process starting and the timeout of
the agent startup script executor. I'm taking the 'lazy' route here and
increasing the timeout to 100ms. This does technically mean that this
makes the test 100 times longer to execute. However, if it takes more
than 100ms to run a `sleep infinity` command on our test runner, I think
we have other issues.
Fixes https://github.com/coder/coder/issues/17063
I'm ignoring flake.nix for now.
```
$ IGNORE_NIX=true ./scripts/check_go_versions.sh
INFO : go.mod : 1.24.1
INFO : dogfood/coder/Dockerfile : 1.24.1
INFO : setup-go/action.yaml : 1.24.1
INFO : flake.nix : 1.22
INFO : Ignoring flake.nix, as IGNORE_NIX=true
Go version check passed, all versions are 1.24.1
$ ./scripts/check_go_versions.sh
INFO : go.mod : 1.24.1
INFO : dogfood/coder/Dockerfile : 1.24.1
INFO : setup-go/action.yaml : 1.24.1
INFO : flake.nix : 1.22
ERROR: Go version mismatch between go.mod and flake.nix
```
Adds a `coder exp mcp` command which will start a local MCP server
listening on stdio with the following capabilities:
* Show logged in user (`coder whoami`)
* List workspaces (`coder list`)
* List templates (`coder templates list`)
* Start a workspace (`coder start`)
* Stop a workspace (`coder stop`)
* Fetch a single workspace (no direct CLI analogue)
* Execute a command inside a workspace (`coder exp rpty`)
* Report the status of a task (currently a no-op, pending task support)
This can be tested as follows:
```
# Start a local Coder server.
./scripts/develop.sh
# Start a workspace. Currently, creating workspaces is not supported.
./scripts/coder-dev.sh create -t docker --yes
# Add the MCP to your Claude config.
claude mcp add coder ./scripts/coder-dev.sh exp mcp
# Tell Claude to do something Coder-related. You may need to nudge it to use the tools.
claude 'start a docker workspace and tell me what version of python is installed'
```
* Improves tests for webpush notifications
* Sets subscriber correctly in web push payload (without this,
notifications do not work in Safari)
* NOTE: for now, I'm using the Coder Access URL. Some push messaging
service don't like it when you use a non-HTTPS URL, so dropping a warn
log about this.
* Adds a service worker and context for push notifications
* Adds a button beside "Inbox" to enable / disable push notifications
Notes:
* ✅ Tested in in Firefox and Safari, and Chrome.
* Adds `codersdk.ExperimentWebPush` (`web-push`)
* Adds a `coderd/webpush` package that allows sending native push
notifications via `github.com/SherClockHolmes/webpush-go`
* Adds database tables to store push notification subscriptions.
* Adds an API endpoint that allows users to subscribe/unsubscribe, and
send a test notification (404 without experiment, excluded from API docs)
* Adds server CLI command to regenerate VAPID keys (note: regenerating
the VAPID keypair requires deleting all existing subscriptions)
---------
Co-authored-by: Kyle Carberry <kyle@carberry.com>
https://github.com/coder/coder/pull/17035 updated the Go version in
`go.mod` and in GH actions but not in `dogfood/coder/Dockerfile` or
`flake.nix`.
This updates the Go version to 1.24 in `dogfood/coder/Dockerfile`.
Unfortunately at the time of writing, Go 1.24 is not available in NixOS.
So that will have to wait.
Fixes an oversight in https://github.com/coder/coder/pull/17032
The FE has logic to replace the string `$SESSION_TOKEN` with a
newly-minted session token.
This adds corresponding logic to the `coder open app` command.
Fixes https://github.com/coder/coder/issues/17009
Adds a CLI command `coder open app <workspace> <app-slug>` that allows
opening arbitrary `coder_apps` via the CLI.
Users can optionally specify a region for workspace
applications.
* Improves separation of concerns between `runDockerInspect` and
`convertDockerInspect`: `runDockerInspect` now just runs the command and
returns the output, while `convertDockerInspect` now does all of the
conversion and parsing logic.
* Improves testing of `convertDockerInspect` using real test fixtures.
* Fixes issue where the container port is returned instead of the host
port.
* Updates UI to link to correct host port. Container port is still
displayed in the button text, but the HostIP:HostPort is shown in a
popover.
* Adds stories for workspace agent UI
`devcontainer.metadata` is apparently an array, not an object. Missed
this first time round!
```
error= get container env info:
github.com/coder/coder/v2/agent/reconnectingpty.(*Server).handleConn
/home/runner/work/coder/coder/agent/reconnectingpty/server.go:193
- read devcontainer remoteEnv:
github.com/coder/coder/v2/agent/agentcontainers.EnvInfo
/home/runner/work/coder/coder/agent/agentcontainers/containers_dockercli.go:119
- unmarshal devcontainer.metadata:
github.com/coder/coder/v2/agent/agentcontainers.devcontainerEnv
/home/runner/work/coder/coder/agent/agentcontainers/containers_dockercli.go:189
- json: cannot unmarshal array into Go value of type struct { RemoteEnv map[string]string "json:\"remoteEnv\"" }
```
If you hit the list containers endpoint with no containers running, the
response is different. This uses a mock lister to ensure a consistent
response from the agent endpoint.
`ServeProvisionerDaemonRequest` has had an ID field for quite a while
now.
This field is only used for telemetry purposes; the actual daemon ID is
created upon insertion in the database. There's no reason to set it, and
it's confusing to do so. Deprecating the field and removing references
to it.
Builds on top of https://github.com/coder/coder/pull/16623/ and wires up
the ReconnectingPTY server. This does nothing to wire up the web
terminal yet but the added test demonstrates the functionality working.
Other changes:
* Refactors and moves the `SystemEnvInfo` interface to the
`agent/usershell` package to address follow-up from
https://github.com/coder/coder/pull/16623#discussion_r1967580249
* Marks `usershellinfo.Get` as deprecated. Consumers should use the
`EnvInfoer` interface instead.
---------
Co-authored-by: Mathias Fredriksson <mafredri@gmail.com>
Co-authored-by: Danny Kopping <danny@coder.com>
Fixes https://github.com/coder/coder/issues/16268
- Adds `/api/v2/workspaceagents/:id/containers` coderd endpoint that allows listing containers
visible to the agent. Optional filtering by labels is supported.
- Adds go tools to the `coder-dylib` CI step so we can generate mocks if needed
Relates to https://github.com/coder/internal/issues/329
It's currently unclear where the SIGHUP came from; adding some logging
to make it more clear if it happens again in future.
---------
Co-authored-by: Danny Kopping <danny@coder.com>
Fixes https://github.com/coder/internal/issues/324
We had been using a `testutil.Context` in combination with a separate
`context.WithTimeout()` that smelled iffy to me.
Also reworked part of the second `require.Eventually` loop to pull a job
ID from the first one, and added some more logging to aid future
debugging.
* chore(docs): update docs re workspace tag default values
* chore(coderdenttest): use random name instead of t.Name() in newExternalProvisionerDaemon
* fix(provisioner/terraform/tfparse): allow empty values in coder_workspace_tag defaults
Relates to https://github.com/coder/coder/issues/15977
Adds support for some functions in `tfparse` (only functions that do not
reference local files).
NOTE: for now, I'm importing trivy-iac. If we prefer to avoid a little
dependency, I can do a little copying instead.
This PR switches the order of precedence of workspace tags when posting
a template version.
Previously, user-specified tags in the request could not override those
detected from our parsing of the template file. Now, they can do.
This addresses a customer issue where were attempting to set a workspace
tag via variable.
Note: there is a possible follow-up item here where we could pass in the
workspace tag values from the request into `tfparse` and let it take
those user-specified values into account. This is covered in a separate
test.
The `make -j` output was sometimes emitting non-JSON output from `go
generate`, resulting in errors like the below:
```
++ /home/coder/src/coder/coder/scripts/coder-dev.sh organizations show me -o json
++ jq -r '.[] | select(.is_default) | .name'
parse error: Invalid numeric literal at line 1, column 3
```
This PR modifies `coder-dev.sh` to silence the output of `make` if the
output is not a TTY.
- Adds `testutil.GoleakOptions` and consolidates existing options to
this location
- Pre-emptively adds required ignore for this Dependabot PR to pass CI
https://github.com/coder/coder/pull/16066
Right now, it looks like we only utilise dlclark/regexp2 as a transitive
dependency of charmbracelet/glamour:
```
$ go mod why github.com/dlclark/regexp2
# github.com/dlclark/regexp2
github.com/coder/coder/v2/coderd/render
github.com/charmbracelet/glamour/ansi
github.com/alecthomas/chroma/v2
github.com/dlclark/regexp2
```
The reason for this replacement was to avoid a goroutine leak in unit
tests.
However, I'm currently unable to replace the goroutine leak as we do not
perform any leak detection tests in `coderd/render`.
If required, we can use the workaround detailed in regexp2's README.md:
https://github.com/dlclark/regexp2/blob/v1.11.4/README.md#goroutine-leak-error
* Improves tfparse test coverage to include more parameter types and values
* Adds tests with unrelated parameters that should be ignored by tfparse
* Modifies tfparse to only attempt evaluation of parameters referenced by coder_workspace_tags
Fixes https://github.com/coder/coder/issues/15972
Adds `keep_locally = true` to other templates that use the
`docker_image` resource to prevent the docker provider from attempting
to remove the image upon workspace deletion.
We had set this in some other places (such as the `dogfood` template)
but had not set this consistently in other templates.
Cian Johnston