⠏ Initializing... found the issue

[3453-315h]

hey there today been have a mission with the docker MCP Server, and trying to troubleshoot this error i know the issue a docker related one, here is the log, below this one is another one which is more important and narrows down the issue.

Authenticated via "gemini-api-key".
[DEBUG] [MCP STDERR (MCP_DOCKER)]: - Reading configuration...

• Reading registry from registry.yaml
  [DEBUG] [MCP STDERR (MCP_DOCKER)]: - Reading catalog from [docker-mcp.yaml]
  [DEBUG] [MCP STDERR (MCP_DOCKER)]: - Reading config from config.yaml
• Reading tools from tools.yaml
• Reading secrets [cyreslab-ai-shodan.api_key exa.api_key gitlab.personal_access_token glif.api_token obsidian.api_key ref.REF_API_KEY]
  [DEBUG] [MCP STDERR (MCP_DOCKER)]: - Configuration read in 729.5497ms
  [DEBUG] [MCP STDERR (MCP_DOCKER)]: - Using images:
• mcp/cyreslab-ai-shodan@sha256:f99f2d2d69555aafcc67229ea7b757db57073d501ed1d60beb47049d3f60e08c
• mcp/exa@sha256:61f88e24df5bc379133f97feb7d5865c2531c56759c8a34f8cd88e02bf33867d
• mcp/gitlab@sha256:a1b8571a210a3c8b17b288498d287cd1c3512c10519330ea71ca48e559e78917
• mcp/glif@sha256:3f93ddceeb993ffddcdceb8cae573b3c93d04ebe7fa61abb630d246c516fc99f
  [DEBUG] [MCP STDERR (MCP_DOCKER)]: - mcp/obsidian@sha256:0eba4c05742ad35faeb91eca40b792454d440d86449c9f1b3cb6c387a510651b
• mcp/playwright@sha256:fd43bc74a3ad644c4dc935a45a3f311a3a93fb4d0ef9a752427601dee01bc217
• mcp/ref-tools-mcp@sha256:ee5129e642802f06c25643b0607a6d68066d07ca8db14c9ae69b13954d19e219
  [DEBUG] [MCP STDERR (MCP_DOCKER)]: > Images pulled in 42.2755ms
• Those servers are enabled: cyreslab-ai-shodan, exa, find-a-domain, gitlab, glif, hugging-face, obsidian, playwright, ref
• Listing MCP tools...
  [DEBUG] [MCP STDERR (MCP_DOCKER)]: - Running mcp/ref-tools-mcp with [run --rm -i --init --security-opt no-new-privileges --cpus 1 --memory 2Gb --pull never -l docker-mcp=true -l docker-mcp-tool-type=mcp -l docker-mcp-name=ref -l docker-mcp-transport=stdio -e REF_API_KEY]
  • Running mcp/cyreslab-ai-shodan with [run --rm -i --init --security-opt no-new-privileges --cpus 1 --memory 2Gb --pull never -l docker-mcp=true -l docker-mcp-tool-type=mcp -l docker-mcp-name=cyreslab-ai-shodan -l docker-mcp-transport=stdio -e SHODAN_API_KEY]
  • Running mcp/exa with [run --rm -i --init --security-opt no-new-privileges --cpus 1 --memory 2Gb --pull never -l docker-mcp=true -l docker-mcp-tool-type=mcp -l docker-mcp-name=exa -l docker-mcp-transport=stdio -e EXA_API_KEY]
  • Running mcp/gitlab with [run --rm -i --init --security-opt no-new-privileges --cpus 1 --memory 2Gb --pull never -l docker-mcp=true -l docker-mcp-tool-type=mcp -l docker-mcp-name=gitlab -l docker-mcp-transport=stdio -e GITLAB_PERSONAL_ACCESS_TOKEN]
  • Running mcp/obsidian with [run --rm -i --init --security-opt no-new-privileges --cpus 1 --memory 2Gb --pull never -l docker-mcp=true -l docker-mcp-tool-type=mcp -l docker-mcp-name=obsidian -l docker-mcp-transport=stdio -e OBSIDIAN_API_KEY -e OBSIDIAN_HOST]
  • Running mcp/playwright with [run --rm -i --init --security-opt no-new-privileges --cpus 1 --memory 2Gb --pull never -l docker-mcp=true -l docker-mcp-tool-type=mcp -l docker-mcp-name=playwright -l docker-mcp-transport=stdio]
  • Running mcp/glif with [run --rm -i --init --security-opt no-new-privileges --cpus 1 --memory 2Gb --pull never -l docker-mcp=true -l docker-mcp-tool-type=mcp -l docker-mcp-name=glif -l docker-mcp-transport=stdio -e GLIF_API_TOKEN -e IGNORE_SAVED_GLIFS]
    [DEBUG] [MCP STDERR (MCP_DOCKER)]: > gitlab: (9 tools)
    [DEBUG] [MCP STDERR (MCP_DOCKER)]: > exa: (1 tools)
    [DEBUG] [MCP STDERR (MCP_DOCKER)]: > glif: (10 tools) (1 prompts) (3 resources)
    [DEBUG] [MCP STDERR (MCP_DOCKER)]: > playwright: (21 tools)
    [DEBUG] [MCP STDERR (MCP_DOCKER)]: > cyreslab-ai-shodan: (5 tools) (1 resources)
    [DEBUG] [MCP STDERR (MCP_DOCKER)]: > obsidian: (12 tools)
    [DEBUG] [MCP STDERR (MCP_DOCKER)]: > hugging-face: (9 tools) (4 prompts)
    [DEBUG] [MCP STDERR (MCP_DOCKER)]: > find-a-domain: (2 tools)
    [DEBUG] [MCP STDERR (MCP_DOCKER)]: - Reading configuration...
  • Reading registry from registry.yaml
    [DEBUG] [MCP STDERR (MCP_DOCKER)]: - Reading catalog from [docker-mcp.yaml]
    [DEBUG] [MCP STDERR (MCP_DOCKER)]: - Reading config from config.yaml
  • Reading tools from tools.yaml
  • Reading secrets [cyreslab-ai-shodan.api_key exa.api_key gitlab.personal_access_token glif.api_token obsidian.api_key]
    [DEBUG] [MCP STDERR (MCP_DOCKER)]: - Configuration read in 943.6127ms

╭──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮
│ You are running Gemini CLI in your home directory. It is recommended to run in a project-specific directory. │
╰──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯

⠹ Initializing...

Using: 1 MCP server
╭────────────────────────────────────────────────────────────────

---

the cli waits for the above to initialize, if i disconnect and reconnect the client in docker i below is the main culprit

mcp-add: tool for adding MCP servers to the registry
mcp-remove: tool for removing MCP servers from the registry
mcp-config-set: tool for setting configuration values for MCP servers
code-mode: write code that calls other MCPs directly
mcp-exec: execute tools that exist in the current session
mcp-discover: prompt for learning about dynamic server management

• Starting OAuth notification monitor
• Starting OAuth provider loops...
  [DEBUG] [MCP STDERR (MCP_DOCKER)]: - Watching for configuration updates...

Initialized in 3.2036956s
Start stdio server

• Connecting to OAuth notification stream at http://localhost/notify/notifications/channel/external-oauth
• Failed to connect to OAuth notifications: Get "http://localhost/notify/notifications/channel/external-oauth": dial unix \.\pipe\dockerBackendApiServer: connect: No connection could be made because the target machine actively refused it.
• Client initialized [email protected]
  [DEBUG] [MCP STDERR (MCP_DOCKER)]: - OAuth notification monitor reconnecting...
• Connecting to OAuth notification stream at http://localhost/notify/notifications/channel/external-oauth
  [DEBUG] [MCP STDERR (MCP_DOCKER)]: - Failed to connect to OAuth notifications: Get "http://localhost/notify/notifications/channel/external-oauth": dial unix \.\pipe\dockerBackendApiServer: connect: No connection could be made because the target machine actively refused it.
  [DEBUG] [MCP STDERR (MCP_DOCKER)]: - OAuth notification monitor reconnecting...
• Connecting to OAuth notification stream at http://localhost/notify/notifications/channel/external-oauth
  [DEBUG] [MCP STDERR (MCP_DOCKER)]: - Failed to connect to OAuth notifications: Get "http://localhost/notify/notifications/channel/external-oauth": dial unix \.\pipe\dockerBackendApiServer: connect: No connection could be made because the target machine actively refused it.
  [DEBUG] [MCP STDERR (MCP_DOCKER)]: - OAuth notification monitor reconnecting...
• Connecting to OAuth notification stream at http://localhost/notify/notifications/channel/external-oauth
  [DEBUG] [MCP STDERR (MCP_DOCKER)]: - Failed to connect to OAuth notifications: Get "http://localhost/notify/notifications/channel/external-oauth": dial unix \.\pipe\dockerBackendApiServer: connect: No connection could be made because the target machine actively refused it.

so just FYI my docker is on, the client for gemini cli is connected, i quickly disconnected to see what would happen and reconnect, it allowed the loading to push forward.

so what my cli looks like (its been open the whole time)

[DEBUG] [MCP STDERR (MCP_DOCKER)]: - Failed to connect to OAuth notifications: Get "http://localhost/notify/notifications/channel/external-oauth": dial unix \.\pipe\dockerBackendApiServer: connect: No connection could be made because the target machine actively refused it.
[DEBUG] [MCP STDERR (MCP_DOCKER)]: - OAuth notification monitor reconnecting...

• Connecting to OAuth notification stream at http://localhost/notify/notifications/channel/external-oauth
  [DEBUG] [MCP STDERR (MCP_DOCKER)]: - Failed to connect to OAuth notifications: Get "http://localhost/notify/notifications/channel/external-oauth": dial unix \.\pipe\dockerBackendApiServer: connect: No connection could be made because the target machine actively refused it.
  [DEBUG] [MCP STDERR (MCP_DOCKER)]: - OAuth notification monitor reconnecting...
• Connecting to OAuth notification stream at http://localhost/notify/notifications/channel/external-oauth
  [DEBUG] [MCP STDERR (MCP_DOCKER)]: - Failed to connect to OAuth notifications: Get "http://localhost/notify/notifications/channel/external-oauth": dial unix \.\pipe\dockerBackendApiServer: connect: No connection could be made because the target machine actively refused it.

╭──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮
│ You are running Gemini CLI in your home directory. It is recommended to run in a project-specific directory. │
╰──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯

Using: 1 MCP server
╭──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮
│ > Type your message or @path/to/file │
╰──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
~ User query: '/mcp list' no sandbox (see /docs) auto | 190.2 MB

i can use the cmd to get a list of the MCP servers tags

🟢 MCP_DOCKER - Ready (75 tools, 6 prompts)
Tools:

• browser_click
• browser_close
• browser_console_messages
• browser_drag
• browser_evaluate
• browser_file_upload
• browser_fill_form
• browser_handle_dialog
• browser_hover
• browser_install
• browser_navigate
• browser_navigate_back
• browser_network_requests
• browser_press_key
• browser_resize
• browser_select_option
• browser_snapshot
• browser_tabs
• browser_take_screenshot
• browser_type
• browser_wait_for
• check_domain
• code-mode
• create_branch
• create_issue
• create_merge_request
• create_or_update_file
• create_repository
• dataset_search
• fork_repository
• get_file_contents
• get_host_info
• get_ssl_info
• glif_info
• gr1_qwen_image_fast_generate_image
• hf_doc_fetch
• hf_doc_search
• hf_whoami
• hub_repo_details
• list_featured_glifs
• list_saved_glif_tools
• list_tlds
• mcp-add
• mcp-config-set
• mcp-exec
• mcp-find
• mcp-remove
• model_search
• my_glif_user_info
• my_glifs
• obsidian_append_content
• obsidian_batch_get_file_contents
• obsidian_complex_search
• obsidian_delete_file
• obsidian_get_file_contents
• obsidian_get_periodic_note
• obsidian_get_recent_changes
• obsidian_get_recent_periodic_notes
• obsidian_list_files_in_dir
• obsidian_list_files_in_vault
• obsidian_patch_content
• obsidian_simple_search
• paper_search
• push_files
• remove_all_glif_tools
• remove_glif_tool
• run_glif
• save_glif_as_tool
• scan_network_range
• search_glifs
• search_iot_devices
• search_repositories
• search_shodan
• space_search
• web_search_exa
  Prompts:
• Dataset Details
• mcp-discover
• Model Details
• Paper Summary
• showcase_featured_glif
• User Summary

and i can actually do things,

[3453-315h]

oh and before you say del .gemini i already did this, uninstalled reinstalled a number of times, even rebooted the pc, and redid things.

i also noticed that there is one MCP server in docker Ref - Up-To-Date Docs, if i enable this gemini just sits on ⠴ Initializing... if i remove the mcp it loads so this MCP server might also have some issue in it.

[3453-315h]

jsut fyi can reproduce this ⠏ Initializing... error 100% of the time if i enable the Ref - Up-To-Date Docs

[3453-315h]

note this issue is for the the devs are aware of. that they need to always ask what type of docker mcp servers the person with the issue is having, cause it most likely will be one of the mcp servers. have not encounter with issue with other mcp servers of yet.