Keepstore can store data in object storage compatible with the S3 API, such as Amazon S3, Google Cloud Storage, or Ceph RADOS.
-h2. Configure keepstore
-
-Copy the "access key" and "secret key" to files where they will be accessible to keepstore at startup time.
-
-<notextile>
-<pre><code>~$ <span class="userinput">sudo sh -c 'cat >/etc/arvados/keepstore/aws_s3_access_key.txt <<EOF'
-zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz==
-EOF</span>
-~$ <span class="userinput">sudo chmod 0400 /etc/arvados/keepstore/aws_s3_access_key.txt</span>
-</code></pre>
-</notextile>
-
-Next, edit the @Volumes@ section of the @keepstore.yml@ config file.
-
-h3. Example config for Amazon S3
-
-<pre>
-Volumes:
-- # The volume type, this indicates object storage compatible with the S3 API
- Type: S3
-
- # Storage provider. If blank, uses Amazon S3 by default.
- # See below for example alternate configuration for Google cloud
- # storage.
- Endpoint: ""
-
- # The bucket to use for the backing store.
- Bucket: example-bucket-name
-
- # The region where the bucket is located.
- Region: us-east-1
-
- # The credentials to use to access the bucket.
- AccessKeyFile: /etc/arvados/keepstore/aws_s3_access_key.txt
- SecretKeyFile: /etc/arvados/keepstore/aws_s3_secret_key.txt
-
- # Maximum time to wait making the initial connection to the backend before
- # failing the request.
- ConnectTimeout: 1m0s
-
- # Page size for s3 "list bucket contents" requests
- IndexPageSize: 1000
-
- # True if the region requires a LocationConstraint declaration
- LocationConstraint: false
-
- # Maximum eventual consistency latency
- RaceWindow: 24h0m0s
-
- # If true, do not accept write or trash operations, only reads.
- ReadOnly: false
-
- # Maximum time to wait for a complete response from the backend before
- # failing the request.
- ReadTimeout: 2m0s
-
- # How much replication is performed by the underlying bucket.
- # This is used to inform replication decisions at the Keep layer.
- S3Replication: 2
-
- # Storage classes to associate with this volume. See "Configuring
- # storage classes" in the "Admin" section of doc.arvados.org.
- StorageClasses: null
-
- # Enable deletion (garbage collection) even when TrashLifetime is
- # zero. WARNING: eventual consistency may result in race conditions
- # that can cause data loss. Do not enable this unless you know what
- # you are doing.
- UnsafeDelete: false
-</pre>
-
-Start (or restart) keepstore, and check its log file to confirm it is using the new configuration.
-
-h3. Example config for Google cloud storage
-
-See previous section for documentation of configuration fields.
-
-</pre>
-Volumes:
-- # Example configuration using alternate storage provider
- # Configuration for Google cloud storage
- Endpoint: https://storage.googleapis.com
- Region: ""
-
- AccessKeyFile: /etc/arvados/keepstore/gce_s3_access_key.txt
- SecretKeyFile: /etc/arvados/keepstore/gce_s3_secret_key.txt
- Bucket: example-bucket-name
- ConnectTimeout: 1m0s
- IndexPageSize: 1000
- LocationConstraint: false
- RaceWindow: 24h0m0s
- ReadOnly: false
- ReadTimeout: 2m0s
- S3Replication: 2
- StorageClasses: null
- UnsafeDelete: false
-</pre>
-
-Start (or restart) keepstore, and check its log file to confirm it is using the new configuration.
+Volumes are configured in the @Volumes@ section of the cluster configuration file.
+
+{% include 'assign_volume_uuid' %}
+
+<notextile><pre><code> Volumes:
+ <span class="userinput">ClusterID</span>-nyw5e-<span class="userinput">000000000000000</span>:
+ AccessViaHosts:
+ # This section determines which keepstore servers access the
+ # volume. In this example, keep0 has read/write access, and
+ # keep1 has read-only access.
+ #
+ # If the AccessViaHosts section is empty or omitted, all
+ # keepstore servers will have read/write access to the
+ # volume.
+ "http://<span class="userinput">keep0.ClusterID.example.com</span>:25107": {}
+ "http://<span class="userinput">keep1.ClusterID.example.com</span>:25107": {ReadOnly: true}
+
+ Driver: <span class="userinput">S3</span>
+ DriverParameters:
+ # Bucket name.
+ Bucket: <span class="userinput">example-bucket-name</span>
+
+ # IAM role name to use when retrieving credentials from
+ # instance metadata. It can be omitted, in which case the
+ # role name itself will be retrieved from instance metadata
+ # -- but setting it explicitly may protect you from using
+ # the wrong credentials in the event of an
+ # installation/configuration error.
+ IAMRole: <span class="userinput">""</span>
+
+ # If you are not using an IAM role for authentication,
+ # specify access credentials here instead.
+ AccessKey: <span class="userinput">""</span>
+ SecretKey: <span class="userinput">""</span>
+
+ # Storage provider region. For Google Cloud Storage, use ""
+ # or omit.
+ Region: <span class="userinput">us-east-1a</span>
+
+ # Storage provider endpoint. For Amazon S3, use "" or
+ # omit. For Google Cloud Storage, use
+ # "https://storage.googleapis.com".
+ Endpoint: ""
+
+ # Change to true if the region requires a LocationConstraint
+ # declaration.
+ LocationConstraint: false
+
+ # Use V2 signatures instead of the default V4. Amazon S3
+ # supports V4 signatures in all regions, but this option
+ # might be needed for other S3-compatible services.
+ V2Signature: false
+
+ # Use the AWS S3 Go driver instead of the goamz driver.
+ AlternateDriver: false
+
+ # Requested page size for "list bucket contents" requests.
+ IndexPageSize: 1000
+
+ # Maximum time to wait while making the initial connection
+ # to the backend before failing the request.
+ ConnectTimeout: 1m
+
+ # Maximum time to wait for a complete response from the
+ # backend before failing the request.
+ ReadTimeout: 2m
+
+ # Maximum eventual consistency latency
+ RaceWindow: 24h
+
+ # How much replication is provided by the underlying bucket.
+ # This is used to inform replication decisions at the Keep
+ # layer.
+ Replication: 2
+
+ # If true, do not accept write or trash operations, even if
+ # AccessViaHosts.*.ReadOnly is false.
+ #
+ # If false or omitted, enable write access (subject to
+ # AccessViaHosts.*.ReadOnly, where applicable).
+ ReadOnly: false
+
+ # Storage classes to associate with this volume. See "Storage
+ # classes" in the "Admin" section of doc.arvados.org.
+ StorageClasses: null
+</code></pre></notextile>
+
+Two S3 drivers are available. Historically, Arvados has used the @goamz@ driver to talk to S3-compatible services. More recently, support for the @aws-sdk-go-v2@ driver was added. This driver can be activated by setting the @AlternateDriver@ flag to @true@.
+
+The @aws-sdk-go-v2@ does not support the old S3 v2 signing algorithm. This will not affect interacting with AWS S3, but it might be an issue when Keep is backed by a very old version of a third party S3-compatible service.
+
+The @aws-sdk-go-v2@ driver has faster _single thread_ read and write performance than the @goamz@ driver. Here are some benchmark numbers against AWS S3, as measured in July 2020. They were generated with the @keep-exercise@ tool in an Arvados installation with one dedicated Keepstore node (c5n.2xlarge) and one dedicated node for running @keep-exercise@ (c5n.2xlarge). The Keepstore node was backed by one S3 bucket, in a VPC with an S3 endpoint installed. Versioning, Server Access logging, Static website hosting, Object-level logging and Default encryption were disabled. Object lock, Transfer acceleration and Requester pays were also disabled. There were no Active notifications. Each test consisted of 4 one minute runs, which were averaged into one number. The tests were repeated 3 times, and of those 3 runs, the highest average speed was selected and included in the table below.
+
+table(table table-bordered table-condensed).
+||_. goamz |_. aws-sdk-go-v2 |_. command line|
+|single thread read performance (average)|32.53 MiB/s|79.48 MiB/s|keep-exercise -repeat 4 -run-time 60s -vary-request -use-index -rthreads 1 -wthreads 1|
+|single thread write performance (average)|39.75 MiB/s|41.05 MiB/s|keep-exercise -repeat 4 -run-time 60s -vary-request -use-index -rthreads 1 -wthreads 1|
+
+Because both S3 and Keep are optimized for _aggregate_ througput, the single thread performance is not as important as it may seem at first glance. When using 20 concurrent read or write threads, the numbers from both drivers are more closely aligned:
+
+table(table table-bordered table-condensed).
+||_. goamz |_. aws-sdk-go-v2 |_. command line|
+|20 thread read performance (average)|585.60 MiB/s|898.93 MiB/s|keep-exercise -repeat 4 -run-time 60s -vary-request -use-index -rthreads 20 -wthreads 0|
+|20 thread write performance (average)|610.40 MiB/s|688.25 MiB/s|keep-exercise -repeat 4 -run-time 60s -vary-request -use-index -rthreads 0 -wthreads 20|
projects / arvados.git / blobdiff