Hi, The keystore and truststore certificates are not getting updated in broker, bookie. I am using the official helm charts to deploy pulsar. Image tag : 3.0.2 Chart version: 3.0.2 Please let me know if there is any toggle that I have to enable for the autoupdate of the certificates in truststore and keystore for the pulsar components.

It seems bookkeeper has CPU spike during compaction , is that true ? I noticed bookkeeper pods pretty busy while deleting some indexes . anyway to reduce CPU usage ?

Hi everyone, I'm currently deploying a Go function into a Kubernetes environment, and I’m encountering an issue related to token management. The function, which runs in a pod, consumes and publishes messages using the token located in

/etc/auth/token

. However, regardless of the token I specify during deployment, the function always seems to end up using the admin token. Here’s how I’m creating the function:

/bin/pulsar-admin \
  --admin-url $ADMIN_URL \
  --auth-params $TOKEN \  # <-- No matter which token I define here, the admin token always ends up in the function instance
  --auth-plugin org.apache.pulsar.client.impl.auth.AuthenticationToken functions create \
  --function-config-file $CONFIG_FILE \
  --go $BIN_PATH \
  --name $FUNCTION_NAME \
  --parallelism $PARALLELISM \
  --cpu $CPU \
  --ram $RAM \
  --user-config "$USER_CONFIG_JSON"

I don't want the function to have full admin rights, does anyone know how to specify which token should be used by the Pulsar function? Thanks in advance for your help!

Hey everyone tell me please if its possible to achieve next thing: I want to create a function and attach to input topic patterns. I want this function to go from newest to oldest. And once all topics are processed, I want to proceed with some other logic. If I understand correctly I can attach function from newest to oldest but only for Java functions is it correct? But what about identifying that all topics were processed, as far as I understand there is nothing built-in for that but can I achieve it by analyzing some function stats or topic-consumer stats etc?

Hi all, I have a problem with the golang-client and authorization. I've setup a pulsar instance on k8s, with Authentication based on K8S Service Accounts, which works perfectly fine. I've furthermore enabled Authorization and granted permission "produce" to "systemserviceaccountmyns:mysa", which, according to the docs should be sufficient to send messages from go. I've created a go client, invoked

createProducer

and I'm receiving an

Authorization failed

error where in the broker logs it says:

DEBUG org.apache.pulsar.broker.authorization.AuthorizationService - Topic operation LOOKUP on topic <non-persistent://tenant/ns/topic> is NOT allowed: role = system:serviceaccount:myns:mysa

How do I grant this Lookup permission?

Hi everyone, I've tried to search for a place in pulsar manager where I can see the live messages in the topic flowing. But there is no place to do so. Can we not know the messages in a topic? I know that pulsar is working and transferring messages across. I can see the topic in pulsar manager. But I can't see any related statistics nor messages in that topic

Hi everyone. Happy Friday! I've been seeing the following issue when trying to update a schema that was previously cleaned up.

Caused by: org.apache.pulsar.broker.service.BrokerServiceException$TopicPoliciesCacheNotInitException: Topic policies cache have not init.

Usually, restarting the brokers "resolves" the issue. Did any of you ever face the same problem? Thanks!

Hi everyone, Are transactions taking place through a transaction api? if so that means we can use it in the pulsar node js library too right?

Hi, I had a pulsar standalone running for several months. During this time a single non-partitioned topic received 250 GB of data. Then I've restarted pulsar. And now it already takes several hours to start, disk io and CPU are busy with processes

PulsarStandaloneStarter

, topic stats says bytesInCounter is around 80 GB and slowly grows. What is this process called? Is this compaction? Where should I learn more about it, how to optimize it etc? Thanks in advance

PIP-371 documentation: https://github.com/apache/pulsar/pull/23143 PIP-371 implementation code: https://github.com/apache/pulsar/pull/23194 Mailing List discussion thread: https://lists.apache.org/thread/42hrf8f5r7yno05hrr6fcz6lgf4jr9f7 This is a native implementation of the request-reply model in pulsar. Although at present, there is no support from many people in the email discussion. But I think things still have to have a result. I am sorry that I am just a personal developer. I have not done pulsar-related work in the company, and I have no business scenarios to test. The implementation code is as follows, if you are interested in this request-reply synchronization model, or if you have the usage scenario of this function. Can help with internal testing. This is just a simple implementation, there are many exception handling or the metrics is not perfect. You are still welcome to discuss your questions on the mailing list or pr.

Hey,I needed a small help, I found most of pulsar broker configurations are missing in official apache pulsar helm chart, How can I add those configurations to helm chart ? (for eg : loadbalancersheddingenabled is missing in helm chart )

Hi Team, Anyone who has got MQTT to work with pulsar on k8s? Thanks

👋 Hi everyone! I m developing an app.I use LDAP as an authentication server. İn pulsar documents ı couldn't see any ldap integration part.I m currently using pulsar version 3.2.3.So how can ı integrate that or is there any resource for it.

How the load Balancing impact the publish latency in pulsar? In production we clearly see the correlation b/w topic unloading and spikes in publish latency.

Greetings, I have been trying to throttle message consumption rate at the namesapce level to 90 msg/sec but fo some unknow reason that still go above the set value

bin/pulsar-admin namespaces set-dispatch-rate public/default --msg-dispatch-rate 90 --dispatch-rate-period 1

I also tried setting the pulish rate to 90 but that as well had no effect, please what more can I do to control this @David K

Is this a multi-partition topic?

If this is from a single subscription, then per-subscription throttling might help.

I'm building like an IoT application with one server and 650,000 clients, where each client acts as a consumer. Due to the large message size and as ı read some article pulsar can handle efficiently that kind of to many clients app, I'm using Pulsar instead of MQTT. My Two Strategies: 1. Strategy 1: Create 650,000 topics, with each client having its own unique topic. 2. Strategy 2: Group 50 clients together per topic, resulting in 13,000 topics. Clients in the same group would use

SubscriptionType.SHARED

, and each message would include a client ID in the header. Only the client whose ID matches the one in the message header would process the message. Questions: 1. Which of these strategies is better for managing 650,000 clients? 2. If Pulsar can handle this load, what kind of cluster setup (e.g., number of brokers, hardware specs) would be sufficient to support this system efficiently? 3. Are there any additional considerations or best practices I should keep in mind when designing a Pulsar-based system with this many clients?

📢 Apache Pulsar Release Policy and Supported Versions 📢 We've posted crucial information about our release policy, currently supported versions, and upcoming releases. Please read the full announcement on the mailing list: https://lists.apache.org/thread/b9mojx7277kpxwrkb9z0xz0g0jnn6n1h Key topics covered: • Time-based release policy • Currently supported versions • Upcoming LTS release (4.0) • Community participation in decisions Your attention to this update is appreciated! Mailing list instructions: https://pulsar.apache.org/contact/

Small Suggestion : for a production setup , running pulsar pods on AWS EKS clustter , how many pods, of broker , zookeeper , bookkeeper and proxy is best suited, like what is the configuration, if we have 3 worker nodes (3 ec2 instances) ?

Hey, folks! Help me to understand the question about outstaffing. I want to understand whether it is still relevant or not. I have created an anonymous questionnaire and I would appreciate if you could share your opinion there! https://forms.gle/1MLCph8PN85YXyiw9

Hello, is there a way assign ownership of a topic to a particular broker?

🎗️https://apache-pulsar.slack.com/archives/C5Z4T36F7/p1718830950250099

Guys I'm finding pulsar too slow, am I doing something wrong. Its way slower that using redis pub/sub. Is this normal? Also I wanted infinite backlogs so maybe that is why. Perhaps I can use both. Would this be a normal practice?

Is this considered good?

2024-08-23T23:54:17,817+0000 [main] INFO org.apache.pulsar.testclient.PerformanceProducer - Throughput produc

ed: 7911 msg --- 791.1 msg/s --- 6.2 Mbit/s --- failure 0.0 msg/s --- Latency: mean: 10.466 ms - med: 9.773 - 95pct: 19.366 - 99pct: 25.210 - 99.9pct: 29.276 - 99.99pct: 34.256 - Max: 35.387 2024-08-23T235427,909+0000 [main] INFO org.apache.pulsar.testclient.PerformanceProducer - Throughput produc ed: 17975 msg --- 999.0 msg/s --- 7.8 Mbit/s --- failure 0.0 msg/s --- Latency: mean: 12.479 ms - med: 9.584 - 95pct: 22.151 - 99pct: 100.745 - 99.9pct: 190.713 - 99.99pct: 197.180 - Max: 197.638 2024-08-23T235437,935+0000 [main] INFO org.apache.pulsar.testclient.PerformanceProducer - Throughput produc ed: 27901 msg --- 983.9 msg/s --- 7.7 Mbit/s --- failure 0.0 msg/s --- Latency: mean: 59.449 ms - med: 19.617 - 95pct: 186.689 - 99pct: 210.741 - 99.9pct: 239.322 - 99.99pct: 246.021 - Max: 247.015 2024-08-23T235447,967+0000 [main] INFO org.apache.pulsar.testclient.PerformanceProducer - Throughput produc ed: 31708 msg --- 379.9 msg/s --- 3.0 Mbit/s --- failure 0.0 msg/s --- Latency: mean: 2203.781 ms - med: 2081.271 - 95pct: 6174.047 - 99pct: 6411.775 - 99.9pct: 6445.503 - 99.99pct: 6449.407 - Max: 6449.4 ?

I have developed a web-based web-based GUI management tool. Feel free to use it! https://github.com/gaecfov/pulsar-console😀

Hi everyone, I am using pulsar 3.1.3 version which includes Trino for me to use SQL to query the produced messages But I encounter a problem where Trino unable to query when the schema has changed, anyone face this problem before? For example, the schema for a topic:-

{ userId: "1234" }

And I able to query using Trino

SELECT userId FROM <topic-name>

And then I add a new property to the schema become:-

{ userId: "1234", userName: "Jeff" }

And try to query with:-

SELECT userId, userName FROM <topic-name>

It returns this error:-

Query 20240823_093959_00006_bcmxz failed: Not a valid schema field: test

And I realise starting from 3.2.x thats no more Trino for us to use? Anyone know why and what are the alternative tools

I am using listeners for consumers. Recently started encountering an issue in prod quite frequently where produced messages were not getting processed for infinite until I consumer machine was restarted. Note this happens to any random consumer all other consumers in the same and different subscriptions could process the messages. However, even after configuring ack timeouts, some messages were never getting processed even by different consumers within the subscription. Was it a network issue, why are messages not being sent to a different consumer in the same sub. I dug deeper and found that the messages were getting consumed, they were in the receiver queue, but listenerExecutor was not processing the messages, and since ack timeout is managed by the consumer and a message is added to the tracker inside the listenerExecutor implementation ack timeout fallback never worked. In my conclusion, listenerExecutor has a single thread that either got paused/killed/no longer exists/starved, etc. Ques: • If I am correct isn't this a frequent issue? • if there is no inbuilt fallback in the official client, any suggestions to overcome this?

Hi everyone, I'm encountering issues with configuring Apache Pulsar to authenticate using Keycloak with OpenID Connect (OIDC). The Docker container for Pulsar is running well, but I'm facing problems with authentication. I have set up the following configuration for the Pulsar broker and the Spring Boot application, but I'm running into issues with authentication. Broker Configuration (broker.conf): # Specify the authentication provider authenticationProviders=org.apache.pulsar.broker.authentication.oidc.AuthenticationProviderOpenID # Enable authentication authenticationEnabled=true # Disable authorization (if not needed) authorizationEnabled=false # OpenID Connect configuration oidc.auth.provider.openid.issuerUrl=http://your-keycloak-server/realms/master oidc.auth.provider.openid.clientCredentialsFile=file:///path/to/your/client-credentials.json oidc.auth.provider.openid.audience=account # Allowed token issuers and audiences openIDAllowedTokenIssuers=http://your-keycloak-server/realms/master openIDAllowedAudiences=account # Optional settings openIDRoleClaim=sub openIDRequireIssuersUseHttps=false # Token validation settings authenticationRefreshCheckSeconds=120 openIDAcceptedTimeLeewaySeconds=120 openIDKeyIdCacheMissRefreshSeconds=120 Spring Boot Application Configuration: URL issuerUrl = new URL("http://your-keycloak-server/realms/master"); URL credentialsUrl = new URL("file:///path/to/your/client-credentials.json"); String audience = "account"; System.out.println("PulsarClient initializing..."); // Build Pulsar Client with OAuth2 Authentication return PulsarClient.builder() .serviceUrl("pulsar://your-pulsar-server:6650") .authentication(AuthenticationFactoryOAuth2.clientCredentials(issuerUrl, credentialsUrl, audience)) .operationTimeout(30, TimeUnit.SECONDS) .connectionTimeout(30, TimeUnit.SECONDS) .build(); Credentials JSON File (client-credentials.json): { "client_id": "client_id", "client_secret": "client_secret", "type": "client_credentials", "grant_type": "client_credentials", } Issues Encountered: Authentication Failures: Errors indicating Failed to authenticate the client and Failed to get partitioned topic metadata. JWT Claims Missing: Errors about missing nbf (not before) claim in the JWT. Configuration Problems: Errors such as Required configuration parameter: privateKey and UnsupportedAuthenticationException. Questions: Missing Claims: Why might the JWT tokens issued by Keycloak be missing the nbf claim, and how can this be resolved? Authentication Failures: What could be causing the Failed to authenticate the client errors, and how can I troubleshoot this? Private Key Error: How can I resolve the Required configuration parameter: privateKey issue when using AuthenticationProviderOpenID? Any insights or suggestions would be greatly appreciated. Thank you!