This repository holds the version of Sarama shipped with Beats, which sometimes includes bug fixes that have not yet reached an official Sarama release. The Beats package links against the beats-fork branch.
The current baseline Sarama version is v1.43.3.
The additional patches applied to this version are:
- A fix for exponential backoff when a Kafka broker is down (issue, pull request, upstream pull request)
- Report common unrecoverable connection / authentication errors (issue, pull request)
If a new fix needs to be applied, open a pull request against the beats-fork branch of this repository. The fix should go through standard code review and should include an update to the "Current state" section of the README.md, linking to details of the change.
This fork does not always sync against the latest Sarama release. However, the intent is to stay as close to the current release as makes sense based on known issues and testing constraints.
Updates should always target an explicit release tag. To perform the update, first create your own fork of sarama with the "Fork" icon at the top right, then run the following from the command line:
# Clone the repository locally, pointing at your own fork
git clone [email protected]:[YOUR USERNAME]/sarama.git
# Add links back to the Elastic and IBM repositories so
# we can merge between them, and run git fetch so we can
# access the version tags.
git remote add upstream [email protected]:elastic/sarama.git
git remote add IBM [email protected]:IBM/sarama.git
git fetch --all
# Check out the current Elastic fork and use it as a baseline
# for a new local branch called "version-upgrade"
git checkout upstream/beats-fork
git checkout -b version-upgrade
# Merge the target version into your new branch (replace
# the version tag as appropriate).
# If the new version conflicts with the outstanding bug
# fixes, you will need to resolve those conflicts in this
# step.
git merge v1.26.4
# Push the update back to your fork on github.
git push --set-upstream origin version-upgrade
After this, open a regular pull request as in the previous section, remembering to update README.md to reflect the new baseline version.
After a new fix or update, the Beats repository needs to be updated to point to the new version. You should almost always target the most recent commit in the commits list for the beats-fork branch unless that would disrupt an impending release.
go get github.com/elastic/sarama@beats-fork
make notice
(When backporting to 7.8 or earlier you will also need to run mage vendor and then rerun make notice in the backport branch.)
You can then commit the results and submit a PR against the Beats repository, remembering to backport if appropriate. Pull requests MUST include the commit summary they target in their description, e.g.:
"This PR updates Sarama to the current Elastic fork, targeting the commit:
commit 123456789abcdefdf68d97cd255ebc039d36e88c (HEAD -> beats-fork, upstream/beats-fork)
Author: Someone <[email protected]>
Date: Wed Jun 10 16:19:46 2020 -0400
Apply some sort of fix
"
(This information is technically redundant, but makes it much easier for code reviewers to confirm that the right version is being applied.) See this example PR.
In any Beats branch, you can find the baseline Sarama version and any additional patches by checking the commit hash in the module configuration:
> grep elastic/sarama go.mod
github.com/elastic/sarama v1.19.1-0.20200625133446-b4d980d71f60
Important: the tag prefix on the Sarama fork, "v1.19.1" in this case, comes from internal interaction between git and go modules, and has no human-discernible connection to the deployed version!
To get the real answer, copy the hash at the end of the line (b4d980d71f60 in this example), and check the repository status at that hash by opening https://github.com/elastic/sarama/tree/[hash]. The README.md at the bottom of that page will list the base version and patches that were in effect in its "Current state" section.
If there is no second line when you check go.mod, then the fork is not in use, and the current Sarama version is whatever is referenced by the tag at the end of the Sarama line.
When a mainline Sarama release includes all fixes we need, we should generally prefer to link directly to it instead of using the fork. To do this:
- Submit a PR against the
beats-forkbranch, updating it to the new target version and reverting any remaining custom patches (see the instructions above). This is to make sure the fork is left in a clean state if it needs to be turned on again in the future. - In the Beats repository:
-
Replace the fork with the upsteam repo
gofmt -w -r '"github.com/elastic/sarama" -> "github.com/IBM/sarama"' . -
select the new Sarama version with:
go get github.com/IBM/[email protected] [replace tag with the real target version] -
Reload the linked version by running
mage vendorandmage update -
Commit the results and submit the resulting PR
-
Once this is done, the fork can be reenabled at any time by following instructions in the previous sections to update the fork and reference it from Beats.
Original README.md follows:
Sarama is an MIT-licensed Go client library for Apache Kafka.
- API documentation and examples are available via pkg.go.dev.
- Mocks for testing are available in the mocks subpackage.
- The examples directory contains more elaborate example applications.
- The tools directory contains command line tools that can be useful for testing, diagnostics, and instrumentation.
You might also want to look at the Frequently Asked Questions.
Sarama provides a "2 releases + 2 months" compatibility guarantee: we support the two latest stable releases of Kafka and Go, and we provide a two month grace period for older releases. However, older releases of Kafka are still likely to work.
Sarama follows semantic versioning and provides API stability via the standard Go module version numbering scheme.
A changelog is available here.
- Get started by checking our contribution guidelines.
- Read the Sarama wiki for more technical and design details.
- The Kafka Protocol Specification contains a wealth of useful information.
- For more general issues, there is a google group for Kafka client developers.
- If you have any questions, just ask!