-
Notifications
You must be signed in to change notification settings - Fork 6
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Issue 32: Update master branch with contents of dev branch #33
Conversation
) Import of current code to development branch with all work done so far. Signed-off-by: Claudio Fahey <[email protected]>
Added transaction support for Pravega driver in OpenMessaging Benchmark. Signed-off-by: wenqimou <[email protected]>
Added transaction support for Kafka. Signed-off-by: wenqimou <[email protected]>
Remove P3 Test Driver framework from OpenMessaging Benchmark fork.
…nnection pooling for writers and time windows for readers. Signed-off-by: Raúl Gracia <[email protected]>
Signed-off-by: Raúl Gracia <[email protected]>
Issue 11: Enable stream auto-scaling
Refractor README
Collocate Segment Store and Bookies and set proper config changes. Signed-off-by: wenqimou <[email protected]>
…default_b_dev Fixing issue 18: PravegaConfig.enableTransaction has the default value of true
Increase sleep time to avoid a race condition when a test completes
* Add option to create scope * Add keycloak dependency
) Avoid allocating a new byte buffer for each write for efficiency reasons. Signed-off-by: Raúl Gracia <[email protected]>
Add configuration to enable Tiering in Pulsar. Signed-off-by: Raúl Gracia <[email protected]> Co-authored-by: Malygina <[email protected]>
Signed-off-by: Raúl Gracia <[email protected]>
…e types. Signed-off-by: Raúl Gracia <[email protected]>
Signed-off-by: Raúl Gracia <[email protected]>
Signed-off-by: Raúl Gracia <[email protected]>
9768588
to
2b35ce7
Compare
benchmark-framework/src/main/java/io/openmessaging/benchmark/WorkloadGenerator.java
Show resolved
Hide resolved
benchmark-framework/src/main/java/io/openmessaging/benchmark/worker/LocalWorker.java
Outdated
Show resolved
Hide resolved
private final ExecutorService executor = Executors.newCachedThreadPool(new DefaultThreadFactory("local-worker")); | ||
// private final ExecutorService executor = Executors.newCachedThreadPool(new DefaultThreadFactory("local-worker")); | ||
// private final ExecutorService executor = Executors.newFixedThreadPool(16); | ||
private final ExecutorService executor = new ForkJoinPool(); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We should explain why we changed from newCachedThreadPool to ForkJoinPool. I think I copied this from Pravega Benchmark.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ok, but what was the reason this change was introduced? The fact that Pravega Benchmark did this does not seem a sufficiently strong reason, right? Is this providing any advantage over the existing thread pool?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Unfortunately, I don't remember exactly why I started using ForkJoinPool. I was likely investigating performance differences between OMB and Pravega Benchmark and I found this to work well. I agree that we should better justify the change or revert it. Are you able to run a benchmark with the original executor?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I will need to do that to see if there is any difference.
...-pravega/src/main/java/io/openmessaging/benchmark/driver/pravega/PravegaBenchmarkDriver.java
Show resolved
Hide resolved
Signed-off-by: Raúl Gracia <[email protected]>
…a/pravega/releases/download/v{{ pravegaVersion}}/pravega-{{ pravegaVersion }}.tgz
[updated] claudiofahey/pravega-tester:0.7.0 from 0.5.1 version |
@jingerbread @claudiofahey I don't think that is a good idea to rely on images from personal repos. If we need an image with whatever is in |
It will take some time to get approvals to make pravega-tester official. Let's remove it from the Ansible script for now. It is not too important. Pravega-tester just performs a few writes and reads from Pravega as a sanity check. |
@RaulGracia, currently in the PR |
@jingerbread why have you closed this PR that is not yours and is still under review? |
Sorry, I did it by mistake |
@claudiofahey @jingerbread we can leave at the moment the Pravega tester image as it is, given that it is useful to know if the deployment works. We can address updating that image on a separate issue: #35. |
@RaulGracia, should be added before In README.md for https://github.com/pravega/openmessaging-benchmark/tree/issue-32-cleanup-dev/driver-pravega#running-the-ansible-playbook Now it's in separate Troubleshooting page https://github.com/pravega/openmessaging-benchmark/blob/issue-32-cleanup-dev/driver-pravega/doc/troubleshooting.md But for proposed in README.md terraform and terraform-inventory versions
|
Should io.pravega:pravega-client version in driver-pravega/pom.xml changed from 0.6.2 version to 0.7.1?
To
|
For Apache Maven 3.6.3
failed on driver-pulsar due to
Need to
|
Seems, when Pravega 0.8 will be release, need to change controller.zk.url to controller.zk.connect.uri in templates/controller.config.properties. Change also zookeeper version in vars.yaml to 3.6.1. May be separate task to update this PR for 0.8 can be created?
pravega/pravega@b4cd667
Otherwise, now with latest Pravega commit b4cd667 from 15 Jul - Pravega deployment wont't work.
However, current setting should work with Pravega 0.7.1, which is currently referenced in vars.yaml |
With mentioned changes - deployment was successful for latest Pravega commit b4cd667 from 15 Jul
|
@jingerbread let me answer your comments:
@jingerbread as you have these changes in place already, can you commit them to this PR please? |
Sure, will commit and create separate ticket for 0.8 changes by the MSK end of the day. |
Besides, may be we can fix warnings by adding license headers?
|
Created |
[changed] driver-pravega/README.md: [added] export TF_STATE=./ before ansible deploy step [changed] driver-pravega/doc/build_pravega.md [added] pravegaSrcRemote and pravegaSrc changing for local artifacts comment [added] comment about -Dlicense.skip=true
@jingerbread @claudiofahey I think we can merge this one as the remaining points can be addressed as separate PRs against |
Seem working for 0.7.1, I thing can be merged. Didn't find how to approve this PR.
|
Change log description
This PR contains the changes from
dev
branch to be merged onmaster
to reproduce the experiments to be published:Still, we need to remove all the non-necessary changes to make this fork as close as possible from the official repo. We also need to remove any mention to personal repos and create the necessary images in Pravega dockerhub.