feat: add agent timings #14713

DanielleMaywood · 2024-09-18T09:29:22Z

Adds timing information for workspace agent scripts.

This change adds:

display_name, id columns to workspace_agent_scripts table
workspace_agent_script_timings table

Demo

Extend the default docker template with the following scripts.

resource "coder_script" "say_hi" {
  agent_id           = coder_agent.main.id
  display_name       = "say hi"
  run_on_start       = true
  start_blocks_login = true

  script = <<EOF
    #!/bin/sh

    echo "Hi!"
  EOF
}

resource "coder_script" "say_cron" {
  agent_id     = coder_agent.main.id
  display_name = "say cron"
  cron         = "*/5 * * * *"

  script = <<EOF
    #!/bin/sh

    echo "Cron!"
  EOF
}

resource "coder_script" "say_bye" {
  agent_id     = coder_agent.main.id
  display_name = "say bye"
  run_on_stop  = true

  script = <<EOF
    #!/bin/sh

    echo "Bye!"
  EOF
}

resource "coder_script" "timeout" {
  agent_id           = coder_agent.main.id
  display_name       = "timeout"
  run_on_start       = true
  start_blocks_login = true
  timeout            = 1

  script = <<EOF
    #!/bin/sh

    sleep 10
  EOF
}

resource "coder_script" "exit_1" {
  agent_id           = coder_agent.main.id
  display_name       = "exit 1"
  run_on_start       = true
  start_blocks_login = true

  script = <<EOF
    #!/bin/sh

    exit 1
  EOF
}

Create a workspace, start it, then stop it. This then gives us the following data in our database.

coder=# select * from workspace_agent_script_timings;

              script_id               |          started_at           |           ended_at            | exit_code | stage |    status
--------------------------------------+-------------------------------+-------------------------------+-----------+-------+--------------
 b3887bc1-3d0b-4eb1-bafb-bc4af3dc19eb | 2024-09-23 16:18:18.370221+00 | 2024-09-23 16:18:18.372466+00 |         0 | start | ok
 1ed547c8-65af-43fd-aa72-47d8866c475c | 2024-09-23 16:18:18.372342+00 | 2024-09-23 16:18:18.373628+00 |         1 | start | exit_failure
 9b8f3503-dd25-4f31-9c47-e715336b9cca | 2024-09-23 16:18:18.372369+00 | 2024-09-23 16:18:19.371824+00 |       255 | start | timed_out
 6e75ef8a-7957-41c7-8cf8-622ed2998c35 | 2024-09-23 16:18:18.371779+00 | 2024-09-23 16:18:21.862226+00 |         0 | start | ok
 11238666-8b11-476b-a6d2-110be27acc9d | 2024-09-23 16:18:25.001315+00 | 2024-09-23 16:18:25.002812+00 |         0 | cron  | ok
 a985fc12-cf2a-49a9-ac65-1c3f98d4bea9 | 2024-09-23 16:18:28.861707+00 | 2024-09-23 16:18:28.862898+00 |         0 | stop  | ok

dannykopping

Excellent work!

agent/agentscripts/agentscripts.go

agent/proto/agent.proto

coderd/agentapi/scripts.go

coderd/provisionerdserver/provisionerdserver.go

agent/agentscripts/agentscripts.go

agent/agentscripts/agentscripts_test.go

coderd/database/dbauthz/dbauthz_test.go

coderd/database/dbmem/dbmem.go

agent/agentscripts/agentscripts_test.go

agent/agenttest/client.go

agent/proto/agent_drpc.pb.go

cli/organizationsettings.go

coderd/database/dump.sql

Not sure how this came to be, I do not recall manually changing these files.

mafredri

Nice job piping this through! And sorry for the amount of comments. 😅

agent/agentscripts/agentscripts.go

mafredri · 2024-09-20T11:39:00Z

agent/proto/agent.proto

+    enum Stage {
+        START = 0;
+        STOP = 1;
+        CRON = 2;


I feel there's a slight mismatch in naming here as I have trouble fitting cron under stage, but I also don't have a better suggestion 😅, just an observation.

Yeah it does feel a little strange. Couldn't think of a better way of handling this though.

mafredri · 2024-09-20T11:52:39Z

coderd/database/migrations/000256_workspace_agent_script_timings.up.sql

+    'cron'
+    );
+
+CREATE TABLE workspace_agent_script_timings


I think calling this timings could work but I feel it's primarily a exec_status or exec_result table, it contains timing information which feels like more of a side-effect.

Renaming it would give us more flexibility to store more data in it if needed.

Renaming a table isn't too difficult; we could do this as a follow-up if needed?

coderd/database/migrations/testdata/fixtures/000256_workspace_agent_script_timings.up.sql

coderd/database/migrations/000256_workspace_agent_script_timings.up.sql

coderd/database/queries/workspaceagents.sql

coderd/database/queries/workspacescripts.sql

johnstcn · 2024-09-24T08:52:16Z

coderd/database/migrations/000257_workspace_agent_script_timings.down.sql

+DROP TYPE IF EXISTS workspace_agent_script_timing_status CASCADE;
+DROP TYPE IF EXISTS workspace_agent_script_timing_stage CASCADE;


note: DROP TYPE ... CASCADE is already done for 000246_provisioner_job_timings

johnstcn

I don't have any further blocking comments.

mafredri

Looking great, thanks for your hard work on this PR!

agent/agentscripts/agentscripts.go

coderd/database/migrations/000257_workspace_agent_script_timings.up.sql

dannykopping

Post-merge approval, excellent work! ✅

github-actions bot assigned DanielleMaywood Sep 18, 2024

DanielleMaywood force-pushed the dm-add-agent-timings branch from 3d5cf74 to 41e8085 Compare September 18, 2024 13:03

dannykopping reviewed Sep 18, 2024

View reviewed changes