debian-mirror-gitlab/doc/development/elasticsearch.md

stage	group	info
Data Stores	Global Search	To determine the technical writer assigned to the Stage/Group associated with this page, see https://about.gitlab.com/handbook/product/ux/technical-writing/#assignments

Elasticsearch knowledge

This area is to maintain a compendium of useful information when working with Elasticsearch.

Information on how to enable Elasticsearch and perform the initial indexing is in the Elasticsearch integration documentation.

Deep Dive

In June 2019, Mario de la Ossa hosted a Deep Dive (GitLab team members only: https://gitlab.com/gitlab-org/create-stage/issues/1) on the GitLab Elasticsearch integration to share his domain specific knowledge with anyone who may work in this part of the codebase in the future. You can find the recording on YouTube, and the slides on Google Slides and in PDF. Everything covered in this deep dive was accurate as of GitLab 12.0, and while specific details may have changed since then, it should still serve as a good introduction.

In August 2020, a second Deep Dive was hosted, focusing on GitLab-specific architecture for multi-indices support. The recording on YouTube and the slides are available. Everything covered in this deep dive was accurate as of GitLab 13.3.

Supported Versions

See Version Requirements.

Developers making significant changes to Elasticsearch queries should test their features against all our supported versions.

Setting up development environment

See the Elasticsearch GDK setup instructions

Helpful Rake tasks

• gitlab:elastic:test:index_size: Tells you how much space the current index is using, as well as how many documents are in the index.
• gitlab:elastic:test:index_size_change: Outputs index size, reindexes, and outputs index size again. Useful when testing improvements to indexing size.

Additionally, if you need large repositories or multiple forks for testing, please consider following these instructions

How does it work?

The Elasticsearch integration depends on an external indexer. We ship an indexer written in Go. The user must trigger the initial indexing via a Rake task but, after this is done, GitLab itself will trigger reindexing when required via after_ callbacks on create, update, and destroy that are inherited from /ee/app/models/concerns/elastic/application_versioned_search.rb.

After initial indexing is complete, create, update, and delete operations for all models except projects (see #207494) are tracked in a Redis ZSET. A regular sidekiq-cron ElasticIndexBulkCronWorker processes this queue, updating many Elasticsearch documents at a time with the Bulk Request API.

Search queries are generated by the concerns found in ee/app/models/concerns/elastic. These concerns are also in charge of access control, and have been a historic source of security bugs so please pay close attention to them!

Existing Analyzers/Tokenizers/Filters

These are all defined in ee/lib/elastic/latest/config.rb

Analyzers

path_analyzer

Used when indexing blobs' paths. Uses the path_tokenizer and the lowercase and asciifolding filters.

Please see the path_tokenizer explanation below for an example.

sha_analyzer

Used in blobs and commits. Uses the sha_tokenizer and the lowercase and asciifolding filters.

Please see the sha_tokenizer explanation later below for an example.

code_analyzer

Used when indexing a blob's filename and content. Uses the whitespace tokenizer and the filters: code, lowercase, and asciifolding

The whitespace tokenizer was selected to have more control over how tokens are split. For example the string Foo::bar(4) needs to generate tokens like Foo and bar(4) to be properly searched.

Please see the code filter for an explanation on how tokens are split.

NOTE: The Elasticsearch code_analyzer doesn't account for all code cases.

code_search_analyzer

Not directly used for indexing, but rather used to transform a search input. Uses the whitespace tokenizer and the lowercase and asciifolding filters.

Tokenizers

sha_tokenizer

This is a custom tokenizer that uses the edgeNGram tokenizer to allow SHAs to be searchable by any sub-set of it (minimum of 5 chars).

Example:

240c29dc7e becomes:

• 240c2
• 240c29
• 240c29d
• 240c29dc
• 240c29dc7
• 240c29dc7e

path_tokenizer

This is a custom tokenizer that uses the path_hierarchy tokenizer with reverse: true to allow searches to find paths no matter how much or how little of the path is given as input.

Example:

'/some/path/application.js' becomes:

• '/some/path/application.js'
• 'some/path/application.js'
• 'path/application.js'
• 'application.js'

Filters

code

Uses a Pattern Capture token filter to split tokens into more easily searched versions of themselves.

Patterns:

• "(\\p{Ll}+|\\p{Lu}\\p{Ll}+|\\p{Lu}+)": captures CamelCase and lowerCamelCase strings as separate tokens
• "(\\d+)": extracts digits
• "(?=([\\p{Lu}]+[\\p{L}]+))": captures CamelCase strings recursively. For example: ThisIsATest => [ThisIsATest, IsATest, ATest, Test]
• '"((?:\\"|[^"]|\\")*)"': captures terms inside quotes, removing the quotes
• "'((?:\\'|[^']|\\')*)'": same as above, for single-quotes
• '\.([^.]+)(?=\.|\s|\Z)': separate terms with periods in-between
• '([\p{L}_.-]+)': some common chars in file names to keep the whole filename intact (for example my_file-ñame.txt)
• '([\p{L}\d_]+)': letters, numbers and underscores are the most common tokens in programming. Always capture them greedily regardless of context.

Gotchas

• Searches can have their own analyzers. Remember to check when editing analyzers
• Character filters (as opposed to token filters) always replace the original character, so they're not a good choice as they can hinder exact searches

Zero downtime reindexing with multiple indices

NOTE: This is not applicable yet as multiple indices functionality is not fully implemented.

Currently GitLab can only handle a single version of setting. Any setting/schema changes would require reindexing everything from scratch. Since reindexing can take a long time, this can cause search functionality downtime.

To avoid downtime, GitLab is working to support multiple indices that can function at the same time. Whenever the schema changes, the administrator will be able to create a new index and reindex to it, while searches continue to go to the older, stable index. Any data updates will be forwarded to both indices. Once the new index is ready, an administrator can mark it active, which will direct all searches to it, and remove the old index.

This is also helpful for migrating to new servers, for example, moving to/from AWS.

Currently we are on the process of migrating to this new design. Everything is hardwired to work with one single version for now.

Architecture

The traditional setup, provided by elasticsearch-rails, is to communicate through its internal proxy classes. Developers would write model-specific logic in a module for the model to include in (for example, SnippetsSearch). The __elasticsearch__ methods would return a proxy object, for example:

• Issue.__elasticsearch__ returns an instance of Elasticsearch::Model::Proxy::ClassMethodsProxy
• Issue.first.__elasticsearch__ returns an instance of Elasticsearch::Model::Proxy::InstanceMethodsProxy.

These proxy objects would talk to Elasticsearch server directly (see top half of the diagram).

In the planned new design, each model would have a pair of corresponding sub-classed proxy objects, in which model-specific logic is located. For example, Snippet would have SnippetClassProxy and SnippetInstanceProxy (being subclass of Elasticsearch::Model::Proxy::ClassMethodsProxy and Elasticsearch::Model::Proxy::InstanceMethodsProxy, respectively).

__elasticsearch__ would represent another layer of proxy object, keeping track of multiple actual proxy objects. It would forward method calls to the appropriate index. For example:

• model.__elasticsearch__.search would be forwarded to the one stable index, since it is a read operation.
• model.__elasticsearch__.update_document would be forwarded to all indices, to keep all indices up-to-date.

The global configurations per version are now in the Elastic::(Version)::Config class. You can change mappings there.

Creating new version of schema

NOTE: This is not applicable yet as multiple indices functionality is not fully implemented.

Folders like ee/lib/elastic/v12p1 contain snapshots of search logic from different versions. To keep a continuous Git history, the latest version lives under ee/lib/elastic/latest, but its classes are aliased under an actual version (for example, ee/lib/elastic/v12p3). When referencing these classes, never use the Latest namespace directly, but use the actual version (for example, V12p3).

The version name basically follows the GitLab release version. If setting is changed in 12.3, we will create a new namespace called V12p3 (p stands for "point"). Raise an issue if there is a need to name a version differently.

If the current version is v12p1, and we need to create a new version for v12p3, the steps are as follows:

1. Copy the entire folder of v12p1 as v12p3
2. Change the namespace for files under v12p3 folder from V12p1 to V12p3 (which are still aliased to Latest)
3. Delete v12p1 folder
4. Copy the entire folder of latest as v12p1
5. Change the namespace for files under v12p1 folder from Latest to V12p1
6. Make changes to files under the latest folder as needed

Creating a new Advanced Search migration

This functionality was introduced by #234046.

NOTE: This only supported for indices created with GitLab 13.0 or greater.

In the ee/elastic/migrate/ folder, create a new file with the filename format YYYYMMDDHHMMSS_migration_name.rb. This format is the same for Rails database migrations.

# frozen_string_literal: true

class MigrationName < Elastic::Migration
  # Important: Any updates to the Elastic index mappings must be replicated in the respective
  # configuration files:
  #   - `Elastic::Latest::Config`, for the main index.
  #   - `Elastic::Latest::<Type>Config`, for standalone indices.

  def migrate
  end

  # Check if the migration has completed
  # Return true if completed, otherwise return false
  def completed?
  end
end

Applied migrations are stored in gitlab-#{RAILS_ENV}-migrations index. All migrations not executed are applied by the Elastic::MigrationWorker cron worker sequentially.

To update Elastic index mappings, apply the configuration to the respective files:

• For the main index: Elastic::Latest::Config.
• For standalone indices: Elastic::Latest::<Type>Config.

Migrations can be built with a retry limit and have the ability to be failed and marked as halted. Any data or index cleanup needed to support migration retries should be handled within the migration.

Migration helpers

The following migration helpers are available in ee/app/workers/concerns/elastic/:

Elastic::MigrationBackfillHelper

Backfills a specific field in an index. In most cases, the mapping for the field should already be added.

Requires the index_name and field_name methods.

class MigrationName < Elastic::Migration
  include Elastic::MigrationBackfillHelper

  private

  def index_name
    Issue.__elasticsearch__.index_name
  end

  def field_name
    :schema_version
  end
end

Elastic::MigrationUpdateMappingsHelper

Updates a mapping in an index by calling put_mapping with the mapping specified.

Requires the index_name and new_mappings methods.

class MigrationName < Elastic::Migration
  include Elastic::MigrationUpdateMappingsHelper

  private

  def index_name
    Issue.__elasticsearch__.index_name
  end

  def new_mappings
    {
      schema_version: {
        type: 'short'
      }
    }
  end
end

Elastic::MigrationRemoveFieldsHelper

Removes specified fields from an index.

Requires the index_name, document_type methods. If there is one field to remove, add the field_to_remove method, otherwise add fields_to_remove with an array of fields.

Checks in batches if any documents that match document_type have the fields specified in Elasticsearch. If documents exist, uses a Painless script to perform update_by_query.

class MigrationName < Elastic::Migration
  include Elastic::MigrationRemoveFieldsHelper

  batched!
  throttle_delay 1.minute

  private

  def index_name
    User.__elasticsearch__.index_name
  end

  def document_type
    'user'
  end

  def fields_to_remove
    %w[two_factor_enabled has_projects]
  end
end

The default batch size is 10_000. You can override this value by specifying BATCH_SIZE:

class MigrationName < Elastic::Migration
  include Elastic::MigrationRemoveFieldsHelper

  batched!
  BATCH_SIZE = 100

  ...
end

Elastic::MigrationObsolete

Marks a migration as obsolete when it's no longer required.

class MigrationName < Elastic::Migration
  include Elastic::MigrationObsolete
end

Elastic::MigrationHelper

Contains methods you can use when a migration doesn't fit the previous examples.

class MigrationName < Elastic::Migration
  include Elastic::MigrationHelper

  def migrate
  ...
  end

  def completed?
  ...
  end
end

Migration options supported by the Elastic::MigrationWorker

Elastic::MigrationWorker supports the following migration options:

• batched! - Allow the migration to run in batches. If set, the Elastic::MigrationWorker will re-enqueue itself with a delay which is set using the throttle_delay option described below. The batching must be handled within the migrate method, this setting controls the re-enqueuing only.

• batch_size - Sets the number of documents modified during a batched! migration run. This size should be set to a value which allows the updates enough time to finish. This can be tuned in combination with the throttle_delay option described below. The batching must be handled within a custom migrate method or by using the Elastic::MigrationBackfillHelper migrate method which uses this setting. Default value is 1000 documents.

• throttle_delay - Sets the wait time in between batch runs. This time should be set high enough to allow each migration batch enough time to finish. Additionally, the time should be less than 30 minutes since that is how often the Elastic::MigrationWorker cron worker runs. Default value is 5 minutes.

• pause_indexing! - Pause indexing while the migration runs. This setting will record the indexing setting before the migration runs and set it back to that value when the migration is completed.

• space_requirements! - Verify that enough free space is available in the cluster when the migration runs. This setting will halt the migration if the storage required is not available when the migration runs. The migration must provide the space required in bytes by defining a space_required_bytes method.

• retry_on_failure - Enable the retry on failure feature. By default, it retries the migration 30 times. After it runs out of retries, the migration is marked as halted. To customize the number of retries, pass the max_attempts argument: retry_on_failure max_attempts: 10

# frozen_string_literal: true

class BatchedMigrationName < Elastic::Migration
  # Declares a migration should be run in batches
  batched!
  throttle_delay 10.minutes
  pause_indexing!
  space_requirements!
  retry_on_failure

  # ...
end

Multi-version compatibility

These Advanced Search migrations, like any other GitLab changes, need to support the case where multiple versions of the application are running at the same time.

Depending on the order of deployment, it's possible that the migration has started or finished and there's still a server running the application code from before the migration. We need to take this into consideration until we can ensure all Advanced Search migrations start after the deployment has finished.

Reverting a migration

Because Elasticsearch does not support transactions, we always need to design our migrations to accommodate a situation where the application code is reverted after the migration has started or after it is finished.

For this reason we generally defer destructive actions (for example, deletions after some data is moved) to a later merge request after the migrations have completed successfully. To be safe, for self-managed customers we should also defer it to another release if there is risk of important data loss.

Best practices for Advanced Search migrations

Follow these best practices for best results:

• When working in batches, keep the batch size under 9,000 documents and throttle_delay for at least 3 minutes. The bulk indexer is set to run every 1 minute and process a batch of 10,000 documents. These limits allow the bulk indexer time to process records before another migration batch is attempted.
• To ensure that document counts are up to date, it is recommended to refresh the index before checking if a migration is completed.
• Add logging statements to each migration when the migration starts, when a completion check occurs, and when the migration is completed. These logs are helpful when debugging issues with migrations.
• Pause indexing if you're using any Elasticsearch Reindex API operations.
• Consider adding a retry limit if there is potential for the migration to fail. This ensures that migrations can be halted if an issue occurs.

Deleting Advanced Search migrations in a major version upgrade

Since our Advanced Search migrations usually require us to support multiple code paths for a long period of time, it's important to clean those up when we safely can.

We choose to use GitLab major version upgrades as a safe time to remove backwards compatibility for indices that have not been fully migrated. We document this in our upgrade documentation. We also choose to replace the migration code with the halted migration and remove tests so that:

• We don't need to maintain any code that is called from our Advanced Search migrations.
• We don't waste CI time running tests for migrations that we don't support anymore.
• Operators who have not run this migration and who upgrade directly to the target version will see a message prompting them to reindex from scratch.

To be extra safe, we will not delete migrations that were created in the last minor version before the major upgrade. So, if we are upgrading to %14.0, we should not delete migrations that were only added in %13.12. This is an extra safety net as we expect there are migrations that get merged that may take multiple weeks to finish on GitLab.com. It would be bad if we upgraded GitLab.com to %14.0 before the migrations in %13.12 were finished. Since our deployments to GitLab.com are automated and we currently don't have automated checks to prevent this, the extra precaution is warranted. Additionally, even if we did have automated checks to prevent it, we wouldn't actually want to hold up GitLab.com deployments on Advanced Search migrations, as they may still have another week to go, and that's too long to block deployments.

Process for removing migrations

For every migration that was created 2 minor versions before the major version being upgraded to, we do the following:

1. Confirm the migration has actually completed successfully for GitLab.com.

2. Replace the content of the migration with:

   include Elastic::MigrationObsolete
   

3. Delete any spec files to support this migration.

4. Remove any logic handling backwards compatibility for this migration. You can find this by looking for Elastic::DataMigrationService.migration_has_finished?(:migration_name_in_lowercase).

5. Create a merge request with these changes. Noting that we should not accidentally merge this before the major release is started.

Performance Monitoring

Prometheus

GitLab exports Prometheus metrics relating to the number of requests and timing for all web/API requests and Sidekiq jobs, which can help diagnose performance trends and compare how Elasticsearch timing is impacting overall performance relative to the time spent doing other things.

Indexing queues

GitLab also exports Prometheus metrics for indexing queues, which can help diagnose performance bottlenecks and determine whether or not your GitLab instance or Elasticsearch server can keep up with the volume of updates.

Logs

All of the indexing happens in Sidekiq, so much of the relevant logs for the Elasticsearch integration can be found in sidekiq.log. In particular, all Sidekiq workers that make requests to Elasticsearch in any way will log the number of requests and time taken querying/writing to Elasticsearch. This can be useful to understand whether or not your cluster is keeping up with indexing.

Searching Elasticsearch is done via ordinary web workers handling requests. Any requests to load a page or make an API request, which then make requests to Elasticsearch, will log the number of requests and the time taken to production_json.log. These logs will also include the time spent on Database and Gitaly requests, which may help to diagnose which part of the search is performing poorly.

There are additional logs specific to Elasticsearch that are sent to elasticsearch.log that may contain information to help diagnose performance issues.

Performance Bar

Elasticsearch requests will be displayed in the Performance Bar, which can be used both locally in development and on any deployed GitLab instance to diagnose poor search performance. This will show the exact queries being made, which is useful to diagnose why a search might be slow.

Correlation ID and X-Opaque-Id

Our correlation ID is forwarded by all requests from Rails to Elasticsearch as the X-Opaque-Id header which allows us to track any tasks in the cluster back the request in GitLab.

Troubleshooting

Getting flood stage disk watermark [95%] exceeded

You might get an error such as

[2018-10-31T15:54:19,762][WARN ][o.e.c.r.a.DiskThresholdMonitor] [pval5Ct]
   flood stage disk watermark [95%] exceeded on
   [pval5Ct7SieH90t5MykM5w][pval5Ct][/usr/local/var/lib/elasticsearch/nodes/0] free: 56.2gb[3%],
   all indices on this node will be marked read-only

This is because you've exceeded the disk space threshold - it thinks you don't have enough disk space left, based on the default 95% threshold.

In addition, the read_only_allow_delete setting will be set to true. It will block indexing, forcemerge, etc

curl "http://localhost:9200/gitlab-development/_settings?pretty"

Add this to your elasticsearch.yml file:

# turn off the disk allocator
cluster.routing.allocation.disk.threshold_enabled: false

or

# set your own limits
cluster.routing.allocation.disk.threshold_enabled: true
cluster.routing.allocation.disk.watermark.flood_stage: 5gb   # ES 6.x only
cluster.routing.allocation.disk.watermark.low: 15gb
cluster.routing.allocation.disk.watermark.high: 10gb

Restart Elasticsearch, and the read_only_allow_delete will clear on its own.

from "Disk-based Shard Allocation | Elasticsearch Reference" 5.6 and 6.x

Disaster recovery/data loss/backups

The use of Elasticsearch in GitLab is only ever as a secondary data store. This means that all of the data stored in Elasticsearch can always be derived again from other data sources, specifically PostgreSQL and Gitaly. Therefore if the Elasticsearch data store is ever corrupted for whatever reason you can reindex everything from scratch.

If your Elasticsearch index is incredibly large it may be too time consuming or cause too much downtime to reindex from scratch. There aren't any built in mechanisms for automatically finding discrepancies and resyncing an Elasticsearch index if it gets out of sync but one tool that may be useful is looking at the logs for all the updates that occurred in a time range you believe may have been missed. This information is very low level and only useful for operators that are familiar with the GitLab codebase. It is documented here in case it is useful for others. The relevant logs that could theoretically be used to figure out what needs to be replayed are:

1. All non-repository updates that were synced can be found in elasticsearch.log by searching for track_items and these can be replayed by sending these items again through ::Elastic::ProcessBookkeepingService.track!
2. All repository updates that occurred can be found in elasticsearch.log by searching for indexing_commit_range. Replaying these requires resetting the IndexStatus#last_commit/last_wiki_commit to the oldest from_sha in the logs and then triggering another index of the project using ElasticCommitIndexerWorker
3. All project deletes that occurred can be found in sidekiq.log by searching for ElasticDeleteProjectWorker. These updates can be replayed by triggering another ElasticDeleteProjectWorker.

With the above methods and taking regular Elasticsearch snapshots we should be able to recover from different kinds of data loss issues in a relatively short period of time compared to indexing everything from scratch.