@@ -72,7 +72,7 @@ Learn how to install, configure, update, and maintain your GitLab instance.
-[Branded login page](../customization/branded_login_page.md): Customize the login page with your own logo, title, and description.
-[Welcome message](../customization/welcome_message.md): Add a custom welcome message to the sign-in page.
-["New Project" page](../customization/new_project_page.md): Customize the text to be displayed on the page that opens whenever your users create a new project.
-[Additional custom email text](https://docs.gitlab.com/ee/user/admin_area/settings/email.html#custom-additional-text): Add additional custom text to emails sent from GitLab. **[PREMIUM ONLY]**
-[Additional custom email text](https://docs.gitlab.com/ee/user/admin_area/settings/email.html#custom-additional-text-premium-only): Add additional custom text to emails sent from GitLab. **[PREMIUM ONLY]**
-[Database](#database-replication): includes the entire application, except cache and jobs.
-[Git repositories](#repository-replication): includes both projects and wikis.
-[Uploaded blobs](#uploads-replication): includes anything from images attached on issues
to raw logs and assets from CI.
With the exception of the Database replication, on a *secondary* node, everything is coordinated
by the [Geo Log Cursor](#geo-log-cursor).
### Geo Log Cursor daemon
The [Geo Log Cursor daemon](#geo-log-cursor-daemon) is a separate process running on
each **secondary** node. It monitors the [Geo Event Log](#geo-event-log)
for new events and creates background jobs for each specific event type.
For example when a repository is updated, the Geo **primary** node creates
a Geo event with an associated repository updated event. The Geo Log Cursor daemon
picks the event up and schedules a `Geo::ProjectSyncWorker` job which will
use the `Geo::RepositorySyncService` and `Geo::WikiSyncService` classes
to update the repository and the wiki respectively.
The Geo Log Cursor daemon can operate in High Availability mode automatically.
The daemon will try to acquire a lock from time to time and once acquired, it
will behave as the *active* daemon.
Any additional running daemons on the same node, will be in standby
mode, ready to resume work if the *active* daemon releases its lock.
We use the [`ExclusiveLease`](https://www.rubydoc.info/github/gitlabhq/gitlabhq/Gitlab/ExclusiveLease) lock type with a small TTL, that is renewed at every
pooling cycle. That allows us to implement this global lock with a timeout.
At the end of the pooling cycle, if the daemon can't renew and/or reacquire
the lock, it switches to standby mode.
### Database replication
Geo uses [streaming replication](#streaming-replication) to replicate
the database from the **primary** to the **secondary** nodes. This
...
...
@@ -13,7 +51,7 @@ replication gives the **secondary** nodes access to all the data saved
in the database. So users can log in on the **secondary** and read all
the issues, merge requests, etc. on the **secondary** node.
## Repository replication
### Repository replication
Geo also replicates repositories. Each **secondary** node keeps track of
the state of every repository in the [tracking database](#tracking-database).
...
...
@@ -23,7 +61,7 @@ There are a few ways a repository gets replicated by the:
@@ -30,7 +30,7 @@ is out of the scope of this document.
Once the data is added to the database or repository and [Elasticsearch is
enabled in the admin area](#enabling-elasticsearch) the search index will be
updated automatically. Elasticsearch can be installed on the same machine as
GitLab, or on a separate server, or you can use the [Amazon Elasticsearch](http://docs.aws.amazon.com/elasticsearch-service/latest/developerguide/es-gsg.html)
GitLab or on a separate server, or you can use the [Amazon Elasticsearch](http://docs.aws.amazon.com/elasticsearch-service/latest/developerguide/es-gsg.html)
service.
You can follow the steps as described in the [official web site](https://www.elastic.co/guide/en/elasticsearch/reference/current/install-elasticsearch.html"Elasticsearch installation documentation") or
...
...
@@ -38,7 +38,7 @@ use the packages that are available for your OS.
## Elasticsearch repository indexer (beta)
In order to improve elasticsearch indexing performance GitLab has made available a [new indexer written in Go](https://gitlab.com/gitlab-org/gitlab-elasticsearch-indexer).
In order to improve elasticsearch indexing performance, GitLab has made available a [new indexer written in Go](https://gitlab.com/gitlab-org/gitlab-elasticsearch-indexer).
This will replace the included Ruby indexer in the future but should be considered beta software for now, so there may be some bugs.
If you would like to use it, please follow the instructions below.
...
...
@@ -56,12 +56,20 @@ installed before running `make`.
@@ -97,10 +107,13 @@ Once installed, enable it under your instance's elasticsearch settings explained
Elasticsearch requires additional resources in excess of those documented in the
[GitLab system requirements](../install/requirements.md). These will vary by
installation size, but you should ensure **at least** an additional 8 GiB of RAM
installation size, but you should ensure **at least** an additional **8 GiB of RAM**
for each Elasticsearch node, per the [official guidelines](https://www.elastic.co/guide/en/elasticsearch/guide/current/hardware.html).
Storage requirements also vary based on installation side, but as a rule of
Keep in mind, this is the **minimum requirements** as per Elasticsearch. For
production instances, they recommend considerably more resources.
Storage requirements also vary based on the installation side, but as a rule of
thumb, you should allocate the total size of your production database, **plus**
two-thirds of the total size of your git repositories. Efforts to reduce this
total are being tracked in this epic: [gitlab-org&153](https://gitlab.com/groups/gitlab-org/-/epics/153).
...
...
@@ -114,7 +127,7 @@ The following Elasticsearch settings are available:
| Parameter | Description |
| --------- | ----------- |
| `Elasticsearch indexing` | Enables/disables Elasticsearch indexing. You may want to enable indexing but disable search in order to give the index time to be fully completed, for example. Also keep in mind that this option doesn't have any impact on existing data, this only enables/disables background indexer which tracks data changes. So by enabling this you will not get your existing data indexed, use special rake task for that as explained in [Adding GitLab's data to the Elasticsearch index](#adding-gitlabs-data-to-the-elasticsearch-index). |
| `Elasticsearch indexing` | Enables/disables Elasticsearch indexing. You may want to enable indexing but disable search in order to give the index time to be fully completed, for example. Also, keep in mind that this option doesn't have any impact on existing data, this only enables/disables background indexer which tracks data changes. So by enabling this you will not get your existing data indexed, use special rake task for that as explained in [Adding GitLab's data to the Elasticsearch index](#adding-gitlabs-data-to-the-elasticsearch-index). |
| `Use the new repository indexer (beta)` | Perform repository indexing using [GitLab Elasticsearch Indexer](https://gitlab.com/gitlab-org/gitlab-elasticsearch-indexer). |
| `Search with Elasticsearch enabled` | Enables/disables using Elasticsearch in search. |
| `URL` | The URL to use for connecting to Elasticsearch. Use a comma-separated list to support clustering (e.g., "http://host1, https://host2:9200"). If your Elasticsearch instance is password protected, pass the `username:password` in the URL (e.g., `http://<username>:<password>@<elastic_host>:9200/`). |
...
...
@@ -152,22 +165,13 @@ To disable the Elasticsearch integration:
1. Find the 'Elasticsearch' section and uncheck 'Search with Elasticsearch enabled'
and 'Elasticsearch indexing'
1. Click **Save** for the changes to take effect
1. [Optional] Delete the existing index by running the command `sudo gitlab-rake gitlab:elastic:delete_index`
## Adding GitLab's data to the Elasticsearch index
### Indexing small instances (database size less than 500 MiB, size of repos less than 5 GiB)
Configure Elasticsearch's host and port in **Admin > Settings > Integrations**. Then create empty indexes using one of the following commands:
- This determines the overall status of the indexing. It is done by counting the total number of indexed projects, dividing by a count of the total number of projects, then multiplying by 100.
Because the `ID_TO` and `ID_FROM` environment variables use the `or equal to` comparison, you can index only one project by using both these variables with the same project ID number:
Indexing project repositories...I, [2019-03-04T21:27:03.083410 #3384] INFO -- : Indexing GitLab User / test (ID=33)...
I, [2019-03-04T21:27:05.215266 #3384] INFO -- : Indexing GitLab User / test (ID=33) is done!
```
## Elasticsearch Index Scopes
When performing a search, the GitLab index will use the following scopes:
| Scope Name | What it searches |
| ---------------- | ---------------------- |
| `commits` | Commit data |
| `projects` | Project data (default) |
| `blobs` | Code |
| `issues` | Issue data |
| `merge_requests` | Merge Request data |
| `milestones` | Milestone data |
| `notes` | Note data |
| `snippets` | Snippet data |
| `wiki_blobs` | Wiki contents |
## Tuning
### Deleted documents
...
...
@@ -357,7 +476,7 @@ However, some larger installations may wish to tune the merge policy settings:
}'
```
- You can also adjust `index.merge.policy.reclaim_deletes_weight`, which controls how aggressively deletions are targetd. But this can lead to costly merge decisions, so we recommend not changing this unless you understand the tradeoffs.
- You can also adjust `index.merge.policy.reclaim_deletes_weight`, which controls how aggressively deletions are targeted. But this can lead to costly merge decisions, so we recommend not changing this unless you understand the tradeoffs.
```bash
curl --request PUT http://localhost:9200/gitlab-production/_settings --data'{
...
...
@@ -373,6 +492,18 @@ However, some larger installations may wish to tune the merge policy settings:
Here are some common pitfalls and how to overcome them:
-**How can I verify my GitLab instance is using Elasticsearch?**
The easiest method is via the rails console (`sudo gitlab-rails console`) by running the following:
```ruby
u = User.find_by_username('your-username')
s = SearchService.new(u, {:search => 'search_term'})
pp s.search_objects.class.name
```
If you see `Elasticsearch::Model::Response::Records`, you are using Elasticsearch.
-**I updated GitLab and now I can't find anything**
We continuously make updates to our indexing strategies and aim to support
...
...
@@ -382,10 +513,26 @@ Here are some common pitfalls and how to overcome them:
-**I indexed all the repositories but I can't find anything**
Make sure you indexed all the database data [as stated above](#adding-gitlabs-data-to-the-elasticsearch-index).
Beyond that, check via the [Elasticsearch Search API](https://www.elastic.co/guide/en/elasticsearch/reference/current/search-search.html) to see if the data shows up on the Elasticsearch side.
If it shows up via the [Elasticsearch Search API](https://www.elastic.co/guide/en/elasticsearch/reference/current/search-search.html), check that it shows up via the rails console (`sudo gitlab-rails console`):
```ruby
u = User.find_by_username('your-username')
s = SearchService.new(u, {:search => 'search_term', :scope => ‘blobs’})
pp s.search_objects.to_a
```
See [Elasticsearch Index Scopes](elasticsearch.md#elasticsearch-index-scopes) for more information on searching for specific types of data.
-**I indexed all the repositories but then switched elastic search servers and now I can't find anything**
You will need to re-run all the rake tasks to re-index the database, repositories, and wikis.
-**The indexing process is taking a very long time**
The more data present in your GitLab instance, the longer the indexing process takes. You might want to try adjusting the BATCH sizes for asynchronous indexing to help speed up the process.
-**No new data is added to the Elasticsearch index when I push code**
@@ -11,37 +11,38 @@ All Geo nodes have the following settings:
| Setting | Description |
| --------| ----------- |
| Primary | This marks a Geo Node as primary. There can be only one primary, make sure that you first add the primary node and then all the others. |
| URL | The instance's full URL, in the same way it is configured in `/etc/gitlab/gitlab.rb` (Omnibus GitLab installations) or `gitlab.yml` (source based installations). |
| Primary | This marks a Geo Node as **primary** node. There can be only one **primary** node; make sure that you first add the **primary** node and then all the others. |
| Name | The unique identifier for the Geo node. Must match the setting `gitlab_rails[geo_node_name]` in `/etc/gitlab/gitlab.rb`. The setting defaults to `external_url` with a trailing slash. |
| URL | The instance's user-facing URL. |
The node you're reading from is indicated with a green `Current node` label, and
the primary is given a blue `Primary` label. Remember that you can only make
changes on the primary!
the **primary** node is given a blue `Primary` label. Remember that you can only make
changes on the **primary** node!
## Secondary node settings
## **Secondary** node settings
Secondaries have a number of additional settings available:
**Secondary** nodes have a number of additional settings available:
| Setting | Description |
|---------------------------|-------------|
Selective synchronization | Enable Geo [selective sync](https://docs.gitlab.com/ee/administration/geo/replication/configuration.html#selective-synchronization) for this **secondary** node. |
| Selective synchronization | Enable Geo [selective sync](https://docs.gitlab.com/ee/administration/geo/replication/configuration.html#selective-synchronization) for this **secondary** node. |
| Repository sync capacity | Number of concurrent requests this **secondary** node will make to the **primary** node when backfilling repositories. |
| File sync capacity | Number of concurrent requests this **secondary** node will make to the **primary** node when backfilling files. |
## Geo backfill
Secondaries are notified of changes to repositories and files by the primary,
**Secondary** nodes are notified of changes to repositories and files by the **primary** node,
and will always attempt to synchronize those changes as quickly as possible.
Backfill is the act of populating the secondary with repositories and files that
existed *before* the secondary was added to the database. Since there may be
Backfill is the act of populating the **secondary** node with repositories and files that
existed *before* the **secondary** node was added to the database. Since there may be
extremely large numbers of repositories and files, it's infeasible to attempt to
download them all at once, so GitLab places an upper limit on the concurrency of
these operations.
How long the backfill takes is a function of the maximum concurrency, but higher
values place more strain on the primary node. From [GitLab 10.2](https://gitlab.com/gitlab-org/gitlab-ee/merge_requests/3107),
the limits are configurable - if your primary node has lots of surplus capacity,
values place more strain on the **primary** node. From [GitLab 10.2](https://gitlab.com/gitlab-org/gitlab-ee/merge_requests/3107),
the limits are configurable. If your **primary** node has lots of surplus capacity,
you can increase the values to complete backfill in a shorter time. If it's
under heavy load and backfill is reducing its availability for normal requests,
you can decrease them.
...
...
@@ -55,3 +56,15 @@ which is used by users. Internal URL does not need to be a private address.
Internal URL defaults to External URL, but you can customize it under
**Admin area > Geo Nodes**.
## Multiple secondary nodes behind a load balancer
In GitLab 11.11, **secondary** nodes can use identical external URLs as long as
a unique `name` is set for each Geo node. The `gitlab.rb` setting
`gitlab_rails[geo_node_name]` must:
- Be set for each GitLab instance that runs `unicorn`, `sidekiq`, or `geo_logcursor`.
- Match a Geo node name.
The load balancer must use sticky sessions in order to avoid authentication
