Skip to content

aitools: look the auto-install gate up once per version, and not without agents - #7756

Merged
anton-107 merged 2 commits into
mainfrom
aitools-gate-lookup-offline
Nov 18, 2026
Merged

anton-107 merged 2 commits into
mainfrom
aitools-gate-lookup-offline

Conversation

@anton-107

@anton-107 anton-107 commented Nov 18, 2026 •

Copy link
Copy Markdown
Contributor

Changes

The auto-install hook resolves its remote gate with one compatibility manifest fetch after a command succeeded. This PR bounds what that fetch costs a machine that cannot reach the manifest:

  • A failed lookup is not repeated for the CLI version (it was repeated on up to 10 commands). The next version looks the gate up again.
  • A machine with no coding agent skips before the lookup. The gate changes nothing for it, and the version is left unsettled so an agent set up later is still found.
  • The lookup is bounded by 1 s instead of 3 s.
  • Ctrl+C or SIGTERM during the lookup ends the lookup instead of the command, which then exits with its own status.

Why

On a network that drops traffic to raw.githubusercontent.com (no internet egress, a firewall, a proxy that never answers), each of the first 10 commands of a version waited 3 s after it had printed its output. A container or job whose home directory is not kept waited on every run, because the attempt count is never stored. A script wrapping a command in timeout 2 failed with 124, and Ctrl+C during the wait turned a successful command into exit 130.

Behavior change to review: with the gate on, a machine without a coding agent no longer reports SKIPPED_NO_AGENTS and no longer settles the version, since it never learns the gate.

No changelog fragment: the lookup has no user-facing entry of its own.

Tests

  • Unit tests: a failed lookup happens once per version and again on the next version; no lookup without agents; a lookup that answers after an interrupt does not install.
  • Acceptance: acceptance/cmd/aitools/auto-install/skipped updated for the no-agents case.
  • By hand, with HTTPS_PROXY pointing at a listener that never answers, three consecutive commands in a fresh home:
before after
a coding agent's config dir present +3.0 s, +3.0 s, +3.0 s +1.0 s, 0, 0
no coding agent +3.0 s, +3.0 s, +3.0 s 0, 0, 0
timeout 2 databricks ... exit 124 (2 of 2 runs) exit 0

SIGINT 0.5 s into the wait: the command exits 0 at once and the next command does not wait.

This pull request and its description were written by Isaac.

…out agents

A machine that cannot reach the compatibility manifest waited for the gate
lookup after each of the first 10 commands of a CLI version, and a machine
whose home directory keeps no state waited on every run.

- A failed lookup is not repeated for the version.
- A machine with no coding agent does not look the gate up.
- The lookup is bounded by 1 s instead of 3 s.
- Ctrl+C or SIGTERM during the lookup ends the lookup, not the command.

Co-authored-by: Isaac <no-reply@databricks.com>
@anton-107
anton-107 requested review from a team and lennartkats-db as code owners November 18, 2026 08:55
@github-actions github-actions Bot added the aitools AI tools (databricks aitools) related changes label Nov 18, 2026
@github-actions

github-actions Bot commented Nov 18, 2026 •

Copy link
Copy Markdown
Contributor

Commit: 9a8c9d819 | Workflow run

Regression test report

Selected 1 changed tests (limit=20, 0 not selected)

1 tests: regression=1 unreleased=0 coverage=0 failing=0
test branch main (0f6d067) latest (v1.26.0)
TestAccept/cmd/aitools/auto-install/skipped ✅ ❌ ❌
TestAccept/cmd/aitools/auto-install/skipped: branch ✅ | main (0f6d067) ❌ | latest (v1.26.0) ❌

main (0f6d067):

acceptance_test.go:1252: Substituted DATABRICKS_CLI_GITHUB_API_URL "$DATABRICKS_HOST" -> "http://127.0.0.1:42517" ("[DATABRICKS_HOST]")
    acceptance_test.go:1252: Substituted DATABRICKS_SKILLS_BASE_URL "$DATABRICKS_HOST" -> "http://127.0.0.1:42517" ("[DATABRICKS_HOST]")
    acceptance_test.go:1384: Diff:
        --- cmd/aitools/auto-install/skipped/output.txt
        +++ /tmp/TestAcceptcmdaitoolsauto-installskipped2617991952/002/output.txt
        @@ -93,3 +93,4 @@
         >>> [CLI] jobs list
         
         >>> print_telemetry
        +{"command":"jobs_list","exit_code":0,"aitools_auto_install_event":{"outcome":"SKIPPED_NO_AGENTS","holdout_arm":"TREATMENT","interactive":false}}
        
    acceptance_test.go:1162: 
        LOG.complete: jobs	The Jobs API allows you to create, edit, and delete jobs.
        LOG.complete: :4
        LOG.complete: Completion ended with directive: ShellCompDirectiveNoFileComp
    acceptance_test.go:1162: 
        LOG.completion: # bash completion V2 for databricks                           -*- shell-script -*-
        LOG.completion: 
        LOG.completion: __databricks_debug()
        LOG.completion: {
        LOG.completion:     if [[ -n ${BASH_COMP_DEBUG_FILE-} ]]; then
        LOG.completion:         echo "$*" >> "${BASH_COMP_DEBUG_FILE}"
        LOG.completion:     fi
        LOG.completion: }
        LOG.completion: 
        LOG.completion: # Macs have bash3 for which the bash-completion package doesn't include
        LOG.completion: # _init_completion. This is a minimal version of that function.
        LOG.completion: __databricks_init_completion()
        LOG.completion: {
        LOG.completion:     COMPREPLY=()
        LOG.completion:     _get_comp_words_by_ref "$@" cur prev words cword
        LOG.completion: }
        LOG.completion: 
        LOG.completion: # This function calls the databricks program to obtain the completion
        LOG.completion: # results and the directive.  It fills the 'out' and 'directive' vars.
        LOG.completion: __databricks_get_completion_results() {
        LOG.completion:     local requestComp lastParam lastChar args
        LOG.completion: 
        LOG.completion:     # Prepare the command to request completions for the program.
        LOG.completion:     # Calling ${words[0]} instead of directly databricks allows handling aliases
        LOG.completion:     args=("${words[@]:1}")
        LOG.completion:     requestComp="${words[0]} __complete ${args[*]}"
        LOG.completion: 
        LOG.completion:     lastParam=${words[$((${#words[@]}-1))]}
        LOG.completion:     lastChar=${lastParam:$((${#lastParam}-1)):1}
        LOG.completion:     __databricks_debug "lastParam ${lastParam}, lastChar ${lastChar}"
        LOG.completion: 
        LOG.completion:     if [[ -z ${cur} && ${lastChar} != = ]]; then
        LOG.completion:         # If the last parameter is complete (there is a space following it)
        LOG.completion:         # We add an extra empty parameter so we can indicate this to the go method.
        LOG.completion:         __databricks_debug "Adding extra empty parameter"
        LOG.completion:         requestComp="${requestComp} ''"
        LOG.completion:     fi
        LOG.completion: 
        LOG.completion:     # When completing a flag with an = (e.g., databricks -n=<TAB>)
        LOG.completion:     # bash focuses on the part after the =, so we need to remove
        LOG.completion:     # the flag part from $cur
        LOG.completion:     if [[ ${cur} == -*=* ]]; then
        LOG.completion:         cur="${cur#*=}"
        LOG.completion:     fi
        LOG.completion: 
        LOG.completion:     __databricks_debug "Calling ${requestComp}"
        LOG.completion:     # Use eval to handle any environment variables and such
        LOG.completion:     out=$(eval "${requestComp}" 2>/dev/null)
        LOG.completion: 
        LOG.completion:     # Extract the directive integer at the very end of the output following a colon (:)
        LOG.completion:     directive=${out##*:}
        LOG.completion:     # Remove the directive
        LOG.completion:     out=${out%:*}
        LOG.completion:     if [[ ${directive} == "${out}" ]]; then
        LOG.completion:         # There is not directive specified
        LOG.completion:         directive=0
        LOG.completion:     fi
        LOG.completion:     __databricks_debug "The completion directive is: ${directive}"
        LOG.completion:     __databricks_debug "The completions are: ${out}"
        LOG.completion: }
        LOG.completion: 
        LOG.completion: __databricks_process_completion_results() {
        LOG.completion:     local shellCompDirectiveError=1
        LOG.completion:     local shellCompDirectiveNoSpace=2
        LOG.completion:     local shellCompDirectiveNoFileComp=4
        LOG.completion:     local shellCompDirectiveFilterFileExt=8
        LOG.completion:     local shellCompDirectiveFilterDirs=16
        LOG.completion:     local shellCompDirectiveKeepOrder=32
        LOG.completion: 
        LOG.completion:     if (((directive & shellCompDirectiveError) != 0)); then
        LOG.completion:         # Error code.  No completion.
        LOG.completion:         __databricks_debug "Received error from custom completion go code"
        LOG.completion:         return
        LOG.completion:     else
        LOG.completion:         if (((directive & shellCompDirectiveNoSpace) != 0)); then
        LOG.completion:             if [[ $(type -t compopt) == builtin ]]; then
        LOG.completion:                 __databricks_debug "Activating no space"
        LOG.completion:                 compopt -o nospace
        LOG.completion:             else
        LOG.completion:                 __databricks_debug "No space directive not supported in this version of bash"
        LOG.completion:             fi
        LOG.completion:         fi
        LOG.completion:         if (((directive & shellCompDirectiveKeepOrder) != 0)); then
        LOG.completion:             if [[ $(type -t compopt) == builtin ]]; then
        LOG.completion:                 # no sort isn't supported for bash less than < 4.4
        LOG.completion:                 if [[ ${BASH_VERSINFO[0]} -lt 4 || ( ${BASH_VERSINFO[0]} -eq 4 && ${BASH_VERSINFO[1]} -lt 4 ) ]]; then
        LOG.completion:                     __databricks_debug "No sort directive not supported in this version of bash"
        LOG.completion:                 else
        LOG.completion:                     __databricks_debug "Activating keep order"
        LOG.completion:                     compopt -o nosort
        LOG.completion:                 fi
        LOG.completion:             else
        LOG.completion:                 __databricks_debug "No sort directive not supported in this version of bash"
        LOG.completion:             fi
        LOG.completion:         fi
        LOG.completion:         if (((directive & shellCompDirectiveNoFileComp) != 0)); then
        LOG.completion:             if [[ $(type -t compopt) == builtin ]]; then
        LOG.completion:                 __databricks_debug "Activating no file completion"
        LOG.completion:                 compopt +o default
        LOG.completion:             else
        LOG.completion: ... (326 more lines truncated, raise DATABRICKS_CLI_TEST_MAX_LOG to see more)
    acceptance_test.go:1162: 
        LOG.help: The Jobs API allows you to create, edit, and delete jobs.
        LOG.help: 
        LOG.help:   You can use a Databricks job to run a data processing or data analysis task in
        LOG.help:   a Databricks cluster with scalable resources. Your job can consist of a single
        LOG.help:   task or can be a large, multi-task workflow with complex dependencies.
        LOG.help:   Databricks manages the task orchestration, cluster management, monitoring, and
        LOG.help:   error reporting for all of your jobs. You can run your jobs immediately or
        LOG.help:   periodically through an easy-to-use scheduling system. You can implement job
        LOG.help:   tasks using notebooks, JARS, Spark Declarative Pipelines, or Python, Scala,
        LOG.help:   Spark submit, and Java applications.
        LOG.help: 
        LOG.help:   You should never hard code secrets or store them in plain text. Use the
        LOG.help:   [Secrets CLI] to manage secrets in the [Databricks CLI]. Use the [Secrets
        LOG.help:   utility] to reference secrets in notebooks and jobs.
        LOG.help: 
        LOG.help:   [Databricks CLI]: https://docs.databricks.com/dev-tools/cli/index.html
        LOG.help:   [Secrets CLI]: https://docs.databricks.com/dev-tools/cli/secrets-cli.html
        LOG.help:   [Secrets utility]: https://docs.databricks.com/dev-tools/databricks-utils.html#dbutils-secrets
        LOG.help: 
        LOG.help: Usage:
        LOG.help:   databricks jobs [flags]
        LOG.help:   databricks jobs [command]
        LOG.help: 
        LOG.help: Main Commands
        LOG.help:   cancel-all-runs       Cancel all runs of a job.
        LOG.help:   cancel-run            Cancel a run.
        LOG.help:   create                Create a new job.
        LOG.help:   delete                Delete a job.
        LOG.help:   delete-run            Delete a job run.
        LOG.help:   export-run            Export and retrieve a job run.
        LOG.help:   get                   Get a single job.
        LOG.help:   get-run               Get a single job run.
        LOG.help:   get-run-output        Get the output for a single run.
        LOG.help:   list                  List jobs.
        LOG.help:   list-runs             List job runs.
        LOG.help:   repair-run            Repair a job run.
        LOG.help:   reset                 Update all job settings (reset).
        LOG.help:   run-now               Trigger a new job run.
        LOG.help:   submit                Create and trigger a one-time run.
        LOG.help:   update                Update job settings partially.
        LOG.help: 
        LOG.help: Permission Commands
        LOG.help:   get-permission-levels Get job permission levels.
        LOG.help:   get-permissions       Get job permissions.
        LOG.help:   set-permissions       Set job permissions.
        LOG.help:   update-permissions    Update job permissions.
        LOG.help: 
        LOG.help: Flags:
        LOG.help:   -h, --help[=true|false]   help for jobs
        LOG.help: 
        LOG.help: Global Flags:
        LOG.help:       --debug[=true|false]   enable debug logging
        LOG.help:   -o, --output type          output type: text or json (default text)
        LOG.help:   -p, --profile string       ~/.databrickscfg profile
        LOG.help:   -t, --target string        bundle target to use (if applicable)
        LOG.help: 
        LOG.help: Use "databricks jobs [command] --help" for more information about a command.
FAIL

latest (v1.26.0):

acceptance_test.go:1252: Substituted DATABRICKS_CLI_GITHUB_API_URL "$DATABRICKS_HOST" -> "http://127.0.0.1:41957" ("[DATABRICKS_HOST]")
    acceptance_test.go:1252: Substituted DATABRICKS_SKILLS_BASE_URL "$DATABRICKS_HOST" -> "http://127.0.0.1:41957" ("[DATABRICKS_HOST]")
    acceptance_test.go:1384: Diff:
        --- cmd/aitools/auto-install/skipped/output.txt
        +++ /tmp/TestAcceptcmdaitoolsauto-installskipped3202070093/002/output.txt
        @@ -4,7 +4,6 @@
         >>> CI=true [CLI] jobs list
         
         >>> print_telemetry
        -{"command":"jobs_list","exit_code":0,"aitools_auto_install_event":{"outcome":"SKIPPED_CI","holdout_arm":"TREATMENT","interactive":false}}
         
         >>> CI=true [CLI] jobs list
         
        @@ -18,32 +17,111 @@
         >>> DATABRICKS_AITOOLS_AUTO_INSTALL=false [CLI] jobs list
         
         >>> print_telemetry
        -{"command":"jobs_list","exit_code":0,"aitools_auto_install_event":{"outcome":"SKIPPED_OPTOUT","holdout_arm":"TREATMENT","interactive":false}}
         
         === auto-install disable: not installed
         
         >>> [CLI] aitools auto-install disable
        -Automatic installation of Databricks AI tools is disabled.
        +Install Databricks skills and plugins into your coding agent so it can work
        +effectively with Databricks resources (bundles, jobs, SQL, and more).
         
        +Supported agents: Claude Code, Cursor, Codex CLI, OpenCode, GitHub Copilot, Antigravity, Pi, Gemini CLI, Goose, Kiro.
        +
        +Skills and plugins are sourced from
        +https://github.com/databricks/databricks-agent-skills
        +
        +Usage:
        +  databricks aitools [command]
        +
        +Available Commands:
        +  install     Install Databricks skills and plugins for coding agents
        +  list        List installed skills and plugins
        +  uninstall   Uninstall Databricks skills and plugins
        +  update      Update installed skills and plugins
        +  version     Show installed skills and plugins version
        +
        +Flags:
        +  -h, --help   help for aitools
        +
        +Global Flags:
        +      --debug            enable debug logging
        +  -o, --output type      output type: text or json (default text)
        +  -p, --profile string   ~/.databrickscfg profile
        +  -t, --target string    bundle target to use (if applicable)
        +
        +Use "databricks aitools [command] --help" for more information about a command.
        +
         >>> [CLI] jobs list
         
         >>> print_telemetry
        -{"command":"jobs_list","exit_code":0,"aitools_auto_install_event":{"outcome":"SKIPPED_OPTOUT","holdout_arm":"TREATMENT","interactive":false}}
         
         >>> cursor_skills
         (no skills)
         
         >>> [CLI] aitools auto-install enable
        -Automatic installation of Databricks AI tools is enabled.
        +Install Databricks skills and plugins into your coding agent so it can work
        +effectively with Databricks resources (bundles, jobs, SQL, and more).
         
        +Supported agents: Claude Code, Cursor, Codex CLI, OpenCode, GitHub Copilot, Antigravity, Pi, Gemini CLI, Goose, Kiro.
        +
        +Skills and plugins are sourced from
        +https://github.com/databricks/databricks-agent-skills
        +
        +Usage:
        +  databricks aitools [command]
        +
        +Available Commands:
        +  install     Install Databricks skills and plugins for coding agents
        +  list        List installed skills and plugins
        +  uninstall   Uninstall Databricks skills and plugins
        +  update      Update installed skills and plugins
        +  version     Show installed skills and plugins version
        +
        +Flags:
        +  -h, --help   help for aitools
        +
        +Global Flags:
        +      --debug            enable debug logging
        +  -o, --output type      output type: text or json (default text)
        +  -p, --profile string   ~/.databrickscfg profile
        +  -t, --target string    bundle target to use (if applicable)
        +
        +Use "databricks aitools [command] --help" for more information about a command.
        +
         === commands that never trigger it
         
         >>> [CLI] version
         Databricks CLI v[CLI_VERSION]
         
         >>> [CLI] aitools auto-install enable
        -Automatic installation of Databricks AI tools is enabled.
        +Install Databricks skills and plugins into your coding agent so it can work
        +effectively with Databricks resources (bundles, jobs, SQL, and more).
         
        +Supported agents: Claude Code, Cursor, Codex CLI, OpenCode, GitHub Copilot, Antigravity, Pi, Gemini CLI, Goose, Kiro.
        +
        +Skills and plugins are sourced from
        +https://github.com/databricks/databricks-agent-skills
        +
        +Usage:
        +  databricks aitools [command]
        +
        +Available Commands:
        +  install     Install Databricks skills and plugins for coding agents
        +  list        List installed skills and plugins
        +  uninstall   Uninstall Databricks skills and plugins
        +  update      Update installed skills and plugins
        +  version     Show installed skills and plugins version
        +
        +Flags:
        +  -h, --help   help for aitools
        +
        +Global Flags:
        +      --debug            enable debug logging
        +  -o, --output type      output type: text or json (default text)
        +  -p, --profile string   ~/.databrickscfg profile
        +  -t, --target string    bundle target to use (if applicable)
        +
        +Use "databricks aitools [command] --help" for more information about a command.
        +
         >>> print_telemetry
         
         >>> DATABRICKS_CACHE_ENABLED=false [CLI] jobs list
        @@ -51,10 +108,7 @@
         >>> print_telemetry
         
         >>> auto_install_state
        -{
        -  "last_version": "1.32.0",
        -  "id": "treatment-machine"
        -}
        +{"id": "treatment-machine"}
         
         >>> cursor_skills
         (no skills)
        @@ -66,10 +120,7 @@
         >>> print_telemetry
         
         >>> auto_install_state
        -{
        -  "last_version": "1.33.0",
        -  "id": "treatment-machine"
        -}
        +{"id": "treatment-machine"}
         
         >>> [CLI] jobs list
         
        @@ -83,7 +134,6 @@
         >>> [CLI] jobs list
         
         >>> print_telemetry
        -{"command":"jobs_list","exit_code":0,"aitools_auto_install_event":{"outcome":"SKIPPED_HOLDOUT","holdout_arm":"HOLDOUT","interactive":false}}
         
         >>> cursor_skills
         (no skills)
        
    acceptance_test.go:1162: 
        LOG.complete: jobs	The Jobs API allows you to create, edit, and delete jobs.
        LOG.complete: :4
        LOG.complete: Completion ended with directive: ShellCompDirectiveNoFileComp
    acceptance_test.go:1162: 
        LOG.completion: # bash completion V2 for databricks                           -*- shell-script -*-
        LOG.completion: 
        LOG.completion: __databricks_debug()
        LOG.completion: {
        LOG.completion:     if [[ -n ${BASH_COMP_DEBUG_FILE-} ]]; then
        LOG.completion:         echo "$*" >> "${BASH_COMP_DEBUG_FILE}"
        LOG.completion:     fi
        LOG.completion: }
        LOG.completion: 
        LOG.completion: # Macs have bash3 for which the bash-completion package doesn't include
        LOG.completion: # _init_completion. This is a minimal version of that function.
        LOG.completion: __databricks_init_completion()
        LOG.completion: {
        LOG.completion:     COMPREPLY=()
        LOG.completion:     _get_comp_words_by_ref "$@" cur prev words cword
        LOG.completion: }
        LOG.completion: 
        LOG.completion: # This function calls the databricks program to obtain the completion
        LOG.completion: # results and the directive.  It fills the 'out' and 'directive' vars.
        LOG.completion: __databricks_get_completion_results() {
        LOG.completion:     local requestComp lastParam lastChar args
        LOG.completion: 
        LOG.completion:     # Prepare the command to request completions for the program.
        LOG.completion:     # Calling ${words[0]} instead of directly databricks allows handling aliases
        LOG.completion:     args=("${words[@]:1}")
        LOG.completion:     requestComp="${words[0]} __complete ${args[*]}"
        LOG.completion: 
        LOG.completion:     lastParam=${words[$((${#words[@]}-1))]}
        LOG.completion:     lastChar=${lastParam:$((${#lastParam}-1)):1}
        LOG.completion:     __databricks_debug "lastParam ${lastParam}, lastChar ${lastChar}"
        LOG.completion: 
        LOG.completion:     if [[ -z ${cur} && ${lastChar} != = ]]; then
... (128 more lines truncated, raise DATABRICKS_CLI_TEST_MAX_LOG to see more)

@review-oss-bot review-oss-bot Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good to me — no issues found.

@eng-dev-ecosystem-bot

eng-dev-ecosystem-bot commented Nov 18, 2026 •

Copy link
Copy Markdown
Collaborator

✅ Integration test report

Commit: 9a8c9d8

Run: 44276859624

Env 💚​RECOVERED ✅​pass 🙈​skip Time
💚​ aws linux-2core-8gb 1 272 20 4:26
💚​ aws-windows-latest-4core-16gb 1 274 18 2:51
💚​ azure linux-2core-8gb 1 271 20 3:52
💚​ azure-windows-latest-4core-16gb 1 273 18 3:46
💚​ gcp linux-2core-8gb 1 272 20 3:38
💚​ gcp-windows-latest-4core-16gb 1 274 18 3:41
Test Name aws linux-2core-8gb aws-windows-latest-4core-16gb azure linux-2core-8gb azure-windows-latest-4core-16gb gcp linux-2core-8gb gcp-windows-latest-4core-16gb
💚​ TestAccept 💚​R 💚​R 💚​R 💚​R 💚​R 💚​R
Top 3 slowest tests (at least 2 minutes):
duration env testname
3:41 azure-windows-latest-4core-16gb TestAccept
3:36 gcp-windows-latest-4core-16gb TestAccept
2:47 aws-windows-latest-4core-16gb TestAccept

@simonfaltum simonfaltum left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Approving with the expectation that the cancellation issue below is addressed before merging.


This review was generated with GitHub MCP.

// Ctrl+C or a request to terminate ends the lookup instead of killing
// a command that succeeded.
gateCtx, stop := signal.NotifyContext(gateCtx, os.Interrupt, syscall.SIGTERM)
in.GateOn, err = gateOn(gateCtx, version)

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

With a stale cached manifest containing auto_install: true, Ctrl+C/SIGTERM cancels gateCtx, but clicompat.fetchManifest falls back to the stale cache and returns success. gateOn then returns (true, nil), so this proceeds to a prompt or installation using the uncanceled parent ctx after the user asked it to stop. Please check cancellation before stop()/cancel() and exit auto-install when interrupted, and add a regression test for the stale-cache case.

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Fixed in 9a8c9d8. The signal context is now separate from the timeout, and a lookup that answers after an interrupt is treated like a failed one: the run ends before any prompt or install, and the version gets no other lookup. TestAutoInstallStopsWhenTheGateLookupIsInterrupted covers the case where the lookup returns the gate on after the interrupt, and fails without the fix.

An interrupted lookup could still resolve the gate from a stale cached
manifest and go on to prompt or install.

Co-authored-by: Isaac <no-reply@databricks.com>
@anton-107
anton-107 enabled auto-merge November 18, 2026 09:31
@anton-107
anton-107 disabled auto-merge November 18, 2026 09:36
@anton-107
anton-107 enabled auto-merge November 18, 2026 09:40
@anton-107
anton-107 added this pull request to the merge queue Nov 18, 2026
Merged via the queue into main with commit 30ea3bf Nov 18, 2026
28 checks passed
@anton-107
anton-107 deleted the aitools-gate-lookup-offline branch November 18, 2026 10:04
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

aitools AI tools (databricks aitools) related changes

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants