[Question]: Tree Mode: Action tries to create new confluence file when one already exists

[joshua-janicas]

Hi there!

Sorry that this is labeled as a bug; I think it's more of a user error on my part but there wasn't any other better option to choose from.

I have been reading documentation on the tree mode format:

https://github.com/Telefonica/confluence-tools/tree/main/components/confluence-sync#tree-mode-1

However when I try to run tree mode with the following configuration:

      - name: Sync Data Dictionary to Confluence
        uses: Telefonica/markdown-confluence-sync-action@fb25441e9a643880dd5ce9494e0e848313bc5a8d
        with:
          mode: tree
          docs-dir: './docs/data-dictionary'
          files-pattern: '*.md'
          confluence-url: 'https://ABC.atlassian.net/wiki'
          confluence-root-page-id: '2573500491'
          confluence-space-key: 'ABC'
          # https://github.com/MrRefactoring/confluence.js?tab=readme-ov-file#authentication
          confluence-authentication: |
            {
              "basic": {
                "email": "${{ vars.ATLASSIAN_CLOUD_API_USERNAME }}",
                "apiToken": "${{ secrets.ATLASSIAN_CLOUD_API_APIKEY }}"
              }
            }
          rehype: |
            {
              "codeBlocks": true
            }

The action runs into an error if the confluence file already exists:

Error creating page with title [staging][content] stg_core_db__product_price_fee: Unexpected Error: com.atlassian.confluence.api.service.exceptions.api.BadRequestException: A page with this title already exists: A page already exists with the same TITLE in this space

And the markdown file looks like this

---
title: stg_core_db__product_price_fee
sync_to_confluence: true
---

# stg_core_db__product_price_fee

(documentation...)

I would think that if it found the file, it should be updating/overwriting instead of doing a complete creation. Am I missing something extra that I should be configuring?

Thank you.

Code of Conduct

• I agree to follow this project's Code of Conduct

[javierbrea]

Hi @joshua-janicas, thanks for the detailed report.

You're raising a good point—in tree mode, the action should indeed update existing pages rather than trying to create them. The error message indicates that the page exists but the action is attempting a creation instead of an update, which suggests the action isn't "finding" the existing page.

Possible Causes

There are a few scenarios that could explain this behavior:

1. Page location: The existing page might not be a child (direct or nested) of the configured root-page-id. In tree mode, the library only looks for pages within the subtree under your root page. If the page exists elsewhere in the space, it won't be recognized for update.

2. Title mismatch: The action generates titles based on the folder structure (e.g., [staging][content] stg_core_db__product_price_fee). If the existing page has a slightly different title, it won't match.

3. Page was created manually or by a different sync: If the page was created outside of this action (manually or with a different configuration), it might not be in the expected hierarchy.

To Help Diagnose

Could you help clarify a few things?

1. Where is the existing page located in Confluence? Is it currently a child page under the root page (2573500491), or is it somewhere else in the space?

2. What is the exact title of the existing page in Confluence? Does it match [staging][content] stg_core_db__product_price_fee exactly?

3. How was this page originally created? Was it created manually, by a previous run of this action, or by another tool?

4. What does your folder structure look like under ./docs/data-dictionary? This helps understand how the action is building the page titles and hierarchy.

Key Documentation References

• In tree mode, pages under the root-page-id not present in the sync list will be deleted
• Pages are identified by title within the tree hierarchy
• The root page must exist before sync, and updating the root page itself is not supported
• Title format in tree mode: [Category A][Category B] Page Title

This information will help pinpoint whether the issue is related to page location, title generation, or something else in the configuration.

Thanks!

[joshua-janicas]

Hi Javier, thanks for responding.

I've been playing around and new pages I have been trying to generate are under a root page identified via confluence-root-page-id: '2571501953'. Was still running into the same problem though.

The first time it creates the page, it generates a new page [staging][content] stg_core_db__product_price_fee, like you outlined how the tree formats the pages [Category A][Category B] Page Title. It properly creates the page structure that matches what i have in my repository.

To summarize:

Where is the existing page located in Confluence? Is it currently a child page under the root page (2573500491), or is it somewhere else in the space? - Everything is generated and kept under the root page (I was trying to do some troubleshooting and thought something was going on with the original page). The current root page is confluence-root-page-id: '2571501953' and all the pages I am looking at were generated by the github action.

   - name: Sync Data Dictionary to Confluence
        uses: Telefonica/markdown-confluence-sync-action@fb25441e9a643880dd5ce9494e0e848313bc5a8d
        with:
          mode: tree
          docs-dir: './docs/data-dictionary'
          files-pattern: '*.md'
          confluence-url: 'https://xyz.atlassian.net/wiki'
          confluence-root-page-id: '2571501953'
          confluence-space-key: 'DWG'
          confluence-authentication: |
            {
              "basic": {
                "email": "${{ vars.ATLASSIAN_CLOUD_API_USERNAME }}",
                "apiToken": "${{ secrets.ATLASSIAN_CLOUD_API_APIKEY }}"
              }
            }
          rehype: |
            {
              "codeBlocks": true
            }

What is the exact title of the existing page in Confluence? Does it match [staging][content] stg_core_db__product_price_fee exactly?

In all cases, yes I believe so:

Back to the example of product price fee:

---
title: stg_core_db__product_price_fee
sync_to_confluence: true
---

Is generated into [staging][content] stg_core_db__product_price_fee. I am currently wondering if it's because the GitHub action is appending the folder hierarchy so to speak that it then gets confused/mismatched (stg_core_db__product_price_fee vs [staging][content]stg_core_db__product_price_fee)? Something I can try on the side to see if that makes a difference.

ADDENDUM: That didn't work :(.

---
title: [staging][content]stg_core_db__product_price_fee
sync_to_confluence: true
---

👇🏼

18:51:25:89 [error][markdown-confluence-sync:markdown:doc-tree:/staging:/content:/stg_core_db__product_price_fee.md] /github/workspace/docs/data-dictionary/staging/content/stg_core_db__product_price_fee.md:1:1-4:4: All collection items must start at the same column
(node:1) YAMLWarning: Keys with collection values will be stringified as YAML due to JS Object restrictions. Use mapAsMap: true to avoid this.

How was this page originally created? Was it created manually, by a previous run of this action, or by another tool? - The pages have been created by the github action

What does your folder structure look like under ./docs/data-dictionary? This helps understand how the action is building the page titles and hierarchy.
In the case of where of stg_core_db__product_price_fee, this is the folder structure

One side note; I have not been using index.md for anything as I'd like the GitHub action to just use my folder naming (to make things simpler on Confluence's side). I tried using index.md's before and had no luck with or without it.

[joshua-janicas]

I tried just restricting the pages I wanted to update down to 1 (set all other pages to not sync), to see what would happen

18:55:37:47 [warn][markdown-confluence-sync:markdown:doc-tree:/staging] Path not exist: Index file does not exist in this path /github/workspace/docs/data-dictionary/staging
...
 18:55:37:49 [warn][markdown-confluence-sync:markdown:doc-tree:/staging:/content] Path not exist: Index file does not exist in this path /github/workspace/docs/data-dictionary/staging/content
...
18:55:41:03 [info][markdown-confluence-sync] Converting 3 markdown documents to Confluence pages...
18:55:41:04 [info][markdown-confluence-sync:confluence] Confluence pages to sync: staging, content, stg_core_db__product_price_fee
18:55:43:24 [info][markdown-confluence-sync:confluence] Confluence pages to sync after transformation: staging, [staging] content, [staging][content] stg_core_db__product_price_fee
18:57:46:83 [error][confluence-sync-pages] Error occurs during sync: Error: Error deleting content with id 2569798442: Unexpected Error: Too many requests to database
Error deleting content with id 2570322633: Unexpected Error: Too many requests to database
Error creating page with title [staging][content] stg_core_db__product_price_fee: Unexpected Error: com.atlassian.confluence.api.service.exceptions.api.BadRequestException: A page with this title already exists: A page already exists with the same TITLE in this space
Error: Error deleting content with id 2569798442: Unexpected Error: Too many requests to database
Error deleting content with id 2570322633: Unexpected Error: Too many requests to database
Error creating page with title [staging][content] stg_core_db__product_price_fee: Unexpected Error: com.atlassian.confluence.api.service.exceptions.api.BadRequestException: A page with this title already exists: A page already exists with the same TITLE in this space

2570322633 / 2569798442 are IDs that match under the same folder but for other models I have.

This is a big hint for me which I havent seen in previous runs. Current running theory is that I have 250 odd Confluence pages I need to sync. I wonder if there's too many it's trying at once and then it ERRORs out, then tries to create them out instead. Will continue to experiment.

[joshua-janicas]

Problem Solved

After clearing out my tree again, and just dealing with 1 sync; the stg_core_db__product_price_fee, I created the page without issue. I was also able to solve a separate issue as to why mermaid diagrams weren't generating :P, as it did it on its own now. This led me further to believe it was a rate limiting prtoblem.

Re-running the job now works as well.

[3](https://github.com/abc/elt-pipeline/actions/runs/21878244802/job/63155519445#step:4:84)
19:16:06:38 [info][markdown-confluence-sync] Converting 3 markdown documents to Confluence pages...
19:16:06:38 [info][markdown-confluence-sync:confluence] Confluence pages to sync: staging, content, stg_core_db__product_price_fee
19:16:08:51 [info][markdown-confluence-sync:confluence] Confluence pages to sync after transformation: staging, [staging] content, [staging][content] stg_core_db__product_price_fee
19:16:11:45 [info][confluence-sync-pages] Sync finished

So yeah, the problem was that I was trying to sync too many pages at once, I hit some kind of API limit, and then everything crashed and burned from there 😂 .

Thank you again for your eyes on this!