This is a custom Gradle build cache implementation which uses AWS S3 to store the cache objects.
This plugin is a fork of myniva/gradle-s3-build-cache. Burrunan adds lots of performance, functional, and security features. See v1.0.0 release notes
Key improvements are (as of 2020-07-24):
- Improved security: the plugin requires only
GetObject
permission, and it does not needListObjects
- Improved security: the plugin uses its own configuration properties rather than default AWS credentials by default
- Faster cache population: cache entries are sent as
File
objects rather than buffering to memory - Faster cache retrieval: cache lookup performs one network request rather than two
maximumCachedObjectLength
property to avoid unexpectedly large remote cache transfers- Elaborate details on cache performance
- Cached items contain metadata (elapsed duration, task name, Gradle version) which help to analyze the cache contents
- Version 1.0.0 - Gradle 4.1+
- Gradle configuration cache - Gradle 7.5+, see gradle/gradle#14874
The plugin is stable, so we encourage you to use it in your projects.
Feedback is very welcome. Please open an issue if you find a bug or have an idea for an improvement.
The Gradle build cache needs to be configured on the Settings level. As a first step, add a
dependency to the plugin to your settings.gradle
file. Get the latest version from Gradle plugin portal.
plugins {
id("com.github.burrunan.s3-build-cache") version "1.8.4"
}
The AWS S3 build cache implementation has a few configuration options:
Configuration Key | Description | Mandatory | Default Value |
---|---|---|---|
region |
The AWS region the S3 bucket is located in. | yes | |
bucket |
The name of the AWS S3 bucket where cache objects should be stored. | yes | |
prefix |
Prefix for the S3 entry names | no | cache/ |
maximumCachedObjectLength |
Maximum object size that can be stored and retrieved from the cache | no | 50'000'000 |
isReducedRedundancy |
Whether or not to use reduced redundancy. | no | true |
endpoint |
Alternative S3 compatible endpoint | no | |
forcePathStyle |
Forces S3 client to use path for bucket names instead of subdomains | no | false |
transferAcceleration |
Configures S3 Transfer Acceleration | no | false |
headers |
A map with HTTP headers to be added to each request (nulls are ignored). e.g. [ 'x-header-name': 'header-value' ] |
no | |
awsAccessKeyId |
The AWS access key id | no | getenv("S3_BUILD_CACHE_ACCESS_KEY_ID") |
awsSecretKey |
The AWS secret key | no | getenv("S3_BUILD_CACHE_SECRET_KEY") |
sessionToken |
The AWS sessionToken when you use temporal credentials | no | getenv("S3_BUILD_CACHE_SESSION_TOKEN") |
awsProfile |
The AWS profile to use for authentication | no | getenv("S3_BUILD_CACHE_PROFILE") |
lookupDefaultAwsCredentials |
Configures if DefaultAWSCredentialsProviderChain could be used to lookup credentials |
yes | false |
showStatistics |
Displays statistics on the remote cache performance | Yes | true |
showStatisticsWhenImpactExceeds |
Specifies minimum duration to trigger printing the stats, milliseconds | Yes | 100 |
showStatisticsWhenSavingsExceeds |
Specifies minimum duration to trigger printing the stats, milliseconds | Yes | 100 |
showStatisticsWhenWasteExceeds |
Specifies minimum duration to trigger printing the stats, milliseconds | Yes | 100 |
showStatisticsWhenTransferExceeds |
Specifies minimum transfer size to trigger printing the stats, bytes | Yes | 1010241024 |
kmsKeyId |
The ID of the KMS key to encrypt cache objects in S3. (Using KMS Encryption) | no |
Note: if both awsAccessKeyId
and awsSecretKey
are nullOrBlank
(null
or whitespace only), then anonymous credentials are used.
The buildCache
configuration block might look like this:
Groovy DSL
// This goes to settings.gradle
apply plugin: 'com.github.burrunan.s3-build-cache'
ext.isCiServer = System.getenv().containsKey("CI")
buildCache {
local {
// Local build cache is dangerous as it might produce inconsistent results
// in case developer modifies files while the build is running
enabled = false
}
remote(com.github.burrunan.s3cache.AwsS3BuildCache) {
region = 'eu-west-1'
bucket = 'your-bucket'
prefix = 'cache/'
push = isCiServer
// Credentials will be taken from S3_BUILD_CACHE_... environment variables
// anonymous access will be used if environment variables are missing
}
}
Kotlin DSL:
// This goes to settings.gradle.kts
plugins {
id("com.github.burrunan.s3-build-cache") version "1.8.4"
}
val isCiServer = System.getenv().containsKey("CI")
buildCache {
local {
// Local build cache is dangerous as it might produce inconsistent results
// in case developer modifies files while the build is running
enabled = false
}
remote<com.github.burrunan.s3cache.AwsS3BuildCache> {
region = "eu-west-1"
bucket = "your-bucket"
prefix = "cache/"
push = isCiServer
}
}
More details about configuring the Gradle build cache can be found in the official Gradle documentation.
It is recommended you specify credentials that have limited access to S3 resources, that is why
plugin retrieves credentials from S3_BUILD_CACHE_ACCESS_KEY_ID
, S3_BUILD_CACHE_SECRET_KEY
, and S3_BUILD_CACHE_SESSION_TOKEN
environment variables.
If you want to use AWS default credentials DefaultAWSCredentialsProviderChain
,
then configure lookupDefaultAwsCredentials=true
.
If you want to use a specific AWS profile,
then configure awsProfile="<profile_name>"
.
Note: even with these values set, it will still try S3_BUILD_CACHE_
variables first.
Note: if you use a path prefix (e.g. build-cache
), you might want to configure the permission to that subfolder only.
{
"Version": "2012-10-17",
"Statement": [
{
"Effect": "Allow",
"Action": [
"s3:PutObject"
],
"Resource": [
"arn:aws:s3:::your-bucket/*"
]
}
]
}
If you use a path prefix (e.g. build-cache
), you might want to configure the permission to that subfolder only.
Note: if you don't have enough permissions to access the item, it will be treated as "cache miss".
{
"Version": "2012-10-17",
"Statement": [
{
"Effect": "Allow",
"Action": [
"s3:GetObject"
],
"Resource": [
"arn:aws:s3:::your-bucket/*"
]
}
]
}
It is important to measure the efficiency of the caching, otherwise it might happen the caching increases build time (e.g. it downloads too large artifacts over a slow network or cache misses are too frequent).
Luckily many cache items include the time it took to build the task, so when the item is loaded
from the cache, the time saved can be estimated as original_task_elapsed_time - from_cache_task_elapsed_time
.
The plugin prints cache statistics at the end of the build (you can disable it with showStatistics=false
):
BUILD SUCCESSFUL in 6s
1 actionable task: 1 executed
S3 cache 232ms saved (242ms saved on hits, 10ms wasted on misses), reads: 2, hits: 1, elapsed: 91ms, processed: 477 B
S3 cache writes: 1, elapsed: 121ms, sent to cache: 472 B
S3 reads:
saved: 232ms
– overall estimation of the remote cache impact on the build time Note: this estimation does not account for parallel task execution.242ms saved on hits
– the estimated time saved by the remote build cache (original_task_elapsed_time - from_cache_task_elapsed_time
). Note: this estimation does not account for parallel task execution.saved on hits
means the cache helps, andwasted on hits
means the estimated execution time would be less than the cache load time.10ms wasted on misses
– the amount of time spent forcache misses
reads: 2
– number load requests to the remote cachehits: 1
– number items loaded from the remote cacheelapsed: 233ms
– total time spent on loading items from remote cacheprocessed: 0 B
– number of bytes loaded from the remote cache
S3 writes:
cache writes: 1
– number of store requests to the remote cacheelapsed: 115ms
– time spent uploading items to the remote cachesent to cache: 472 B
– number of bytes sent to the remote cache
The stored cache entries might include metadata. The metadata helps to estimate cache efficiency, and it might be useful to analyze space consumption.
Key | Type | Sample | Description |
---|---|---|---|
buildInvocationId | String | rpha3qmrzvbnxhmdlvukwcx7ru | Build Id |
identity | String | :example:test | Task identifier |
executionTime | Long | 189871 | Task execution time, milliseconds |
operatingSystem | String | Linux | Operating system |
gradleVersion | String | 6.3 | Gradle version |
This plugin does not deal with expiring cache entries directly but relies on S3 object lifecycle management to do so. Cache entry expiration rules can be set on S3 buckets using AWS API or via AWS Management Console.
Contributions are always welcome! If you'd like to contribute (and we hope you do) please open a pull request.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.