Summary

Bitbucket code search does not return the expected results for a specific repository.

The considerations mentioned in the documentation Search for code in Bitbucket Server have been reviewed and the search results are considered incorrect.

For example:

• a string that is expected to be found in a search is not returned
• a string that is not expected to be found in a search is returned

A reindex for the repo can be attempted.

If issues with search are widespread and observed in a lot of repos, please see Bitbucket Server Code Search is unable to find results from some or all repositories.

Environment

This applies to Bitbucket Server version 6.0 and later.

Cause

An error might have occurred at the time the repo is being indexed to the Search server (Elasticsearch or OpenSearch), such as when the instance ran low on disk space or if there were connectivity issues.

Note that indexing to the search server is only attempted up to 5 times. Even if the underlying root cause has been resolved, the indexing process will never try indexing a repo that has reached the maximum retry count again because the number of retries are stored in the index as well.

Before proceeding with the resolution described here, you will first need to ensure that the underlying problem that is causing indexing to fail is resolved. If you need help with this process, contact Atlassian Support via support.atlassian.com. Once the root cause has been found and the problem has been resolved, follow the resolution steps below to reindex a repo.

Solution

Step 1: Check the number of indexing retries

Check whether the number of indexing retries for the repo has already been reached or not by doing the following:

Get the repo id using the steps in Identifying a Repository on the Bitbucket Server

Then call the endpoint:

curl -u <SEARCH_USERNAME>:<SEARCH_PASSWORD> -X GET <SEARCH_SERVER_URL>/bitbucket-index-state/_doc/<REPO_ID>

where:

• SEARCH_USERNAME
  • If you are using Elasticsearch, this is the auth.basic.username value in the buckler.yml file
    If you are using the embedded Elasticsearch, the buckler.yml file is located in <BITBUCKET_HOME>/shared/search/bucker/buckler.yml
    If you have a remote Elasticsearch instance, the buckler.yml file location would depend on how the remote Elasticsearch instance was configured.
  • If you are using OpenSearch, this is the user configured in the internal_users.yml  file 
    See remote OpenSearch server - security configuration for more info
• SEARCH_PASSWORD
  • If you are using Elasticsearch, this is the auth.basic.password value in the buckler.yml file
  • If you are using OpenSearch, this is the password for the user configured in the internal_users.yml  file 
• SEARCH_SERVER_URL is the URL for the search server instance
  You can get this from the Bitbucket UI in: Administration > Server settings > Search > Search server URL
• REPO_ID is the repository id

This will output data on the repo's indexing state.

Sample output:

{
    "_index": "bitbucket-index-state",
    "_type": "_doc",
    "_id": "1",
    "_version": 649,
    "_seq_no": 6798,
    "_primary_term": 302,
    "found": true,
    "_source": {
        "public": false,
        "fork": false,
        "retries": 6,
        "projectId": 1,
        "indexedCommitId": "ec3deb7ae0280892f0689eb5582252c116626183"
    }
}

Look for the "retries" value in the output.

If the "retries" value is 5 or less, you can proceed directly to Step 3: Reindex the repo. 

If the "retries" is above 5, you need to go through Step 2: Reset the indexing retries count if the maximum limit has been reached first.

Step 2: Reset the indexing retries count if the maximum limit has been reached

Reset the "retries"  count to 0 with the following procedures:

Copy the "_source" data from Step 1 and change the "retries" value to 0. For example:

{
        "public": false,
        "fork": false,
        "retries": 0,
        "projectId": 1,
        "indexedCommitId": "ec3deb7ae0280892f0689eb5582252c116626183"
}

Pass this data as the JSON_DATA value to the following PUT call:

curl -u <SEARCH_USERNAME>:<SEARCH_PASSWORD> -X PUT <SEARCH_SERVER_URL>/bitbucket-index-state/_doc/<REPO_ID> \
   -H 'Content-Type: application/json' \
   -d '<JSON_DATA>'

For example:

curl -u es_admin:es_password -X PUT http://localhost:7992/bitbucket-index-state/_doc/1 \
   -H 'Content-Type: application/json' \
   -d '{
        "public": false,
        "fork": false,
        "retries": 0,
        "projectId": 1,
        "indexedCommitId": "ec3deb7ae0280892f0689eb5582252c116626183"
       }'

Step 3: Reindex the repo

Reindex the repo for code search by calling the REST endpoint:

curl -u <USERNAME> -X POST -H 'Content-Type: application/json' -H 'Accept: application/json' <BITBUCKET_URL>/rest/indexing/latest/projects/<PROJECT_KEY>/repos/<REPO_SLUG>/sync

where:

• USERNAME is the username of a Bitbucket Server user with administrative privileges
• BITBUCKET_URL is the base URL for the instance
• PROJECT_KEY is the project key in uppercase
• REPO_SLUG is the repository slug in lowercase

Afterwards, test the search again.