38 results found
-
Create a Backup to Azure Blob Storage for Local Region Restore - DR
As an application owner,
I want to be able to store data in Blob Storage for local recover from an outage using the backups on Blob storage and also be able to restore accidentally dropped topics. This backup would potentially include hundreds of topics.2 votes -
ACL
As SRE Engineer
I want to customize ACL to allow kafka consumer operations to carry on, even while write lock gets triggered when the disk space reaches threshold limits of 95 or 97%. Given that the Kafka consumers' offset commits are relatively smaller, this option will not be detrimental.
so that even when disk space reaches critical levels, it will not immediately impact consumer side operations
1 vote -
Support Grouping of OAuth2/OIDC Users
As a platform engineer
I want to group multiple users based on their role (OAuth2/OIDC claim)
so that I can reduce the number of required Kafka users and ACL entries that need to be managed.Currently, every user / identity connecting via OAuth2/OIDC has a 1:1 mapping to a Kafka user (the username is taken from the sub claim). This is cumbersome and leads to significant overhead if for example multiple identities / users with the same permissions want to access the Kafka service. Kafka users and ACLs need to be created for every single identity, even though they share…
2 votes -
Monitoring consume lag for kafka not out of the box
As a Solutions Architect
I want to have consumer lag for kafka consumers available out of the box
so that I can correctly monitor streaming applications without having to setup an external prometheus. In addition Consumer lag is the key metric to monitor for end to end health of streaming apps(ie to ensure they are keeping up with demand), you cannot put a streaming app into production with out correct monitoring and alerting on this metric.Background
1. there is a consumer lag on the default metrics dashboard but it does not work.
2. I contact support and found that…1 vote -
untyped metrics from prometheus endpoint
using prometheus endpoint at our kafka-service we need to have metrics from kafka with type .
see below the metrics have notype - this should be set at the endpoint ,to ease the use of these metrics .
see exampleTYPE kafkaservergroupcoordinatormetricsgroupcompletedrebalancecount untyped
TYPE kafkaservergroupcoordinatormetricsoffsetcommit_rate untyped
1 vote -
"Last Used" field on Kafka Certificates in Console/API
As an organization (DevOps/Security/Vendor Manager) using Aiven Kafka,
we want to determine the last connected cert status of Kafka users,
so that we can know whether a kafka user certificate has been successfully updated.We automate certificate rolling to an extent with terraform. Different teams of devs generally own their section of terraform creating Kafka users. As of right now every two years those certs expire, and clicking the "Yes I've updated" in the Aiven console just silences the alert, and provides no real time verification from the running kafka that a certificate has been updated.
This means a user…
1 vote -
fetch closest replica
As a Kafka user
I want to be able to have consumers fetch from the closest replica (KIP-392)
so that I can reduce Inter-AZ costs
In addition, this will also reduce latency of consumer calls.2 votes -
Support for both allow and black lists in Kafka service
As a managed Kafka administrator
I want to have the ability to black-list a bunch of IP addresses I believe are suspicious from my Kafka service, so that I can prevent my service being disrupted by unexpected traffic
In addition, I would like to keep the current allow-list to be able to allow know IPs and a way to resolve a conflict between allow and block list where block list takes priority.1 voteShelved ·AdminMichael Tansini (Product Manager Data Streaming (Kafka and Flink), Aiven.io) responded
at the moment we recommend using our current networking whitelisting capabilities
-
Cluster leader balancing CPU vs Disk - can we choose?
As an OPS Engineer
I want to be able to balance the cluster based on CPU usage rather than Disk usage. The current algorithm focuses on Disk usage which is not optimum for our application.
Can we have an option to apply partition rebalancing based on CPU usage?1 vote -
Kafka versioned updates
As a platform engineer
I want the service updates to be versioned and to be able to select a specific version (Kafka, Karapace, etc) to update to so that I can perform correct change promotion from non-prod to production (instead of being forced to always apply the latest) and so that if a new version is released while an update is in-flight it would not lead to different versions running across the brokers (as is currently the case, which we have experienced can lead to incompatibilities). Additionally, I would like it to be clear what version(s) is currently running in…12 votesRoadmapped ·This is partially being realised with users being able to view what service is available and what update will take place.
-
Kafka fined grained ACLs
As an ops engineer,
I want to be able to declare fined grained ACLs
so that I avoid having to grant "admin" rights to users which only need "DeleteRecords" rights on specific topics.
The client uses Kafka Streams, which needs specific rights (https://docs.confluent.io/platform/current/streams/developer-guide/security.html#required-acl-setting-for-secure-ak-clusters) which are not covered by aiven predefined rights.
Currently, the "admin" is to broad for such access (I don't want the user to be able to create topics).8 votes -
Ability to scale Kafka cluster without upgrade
As a Kafka Administrator
I want to ahve ability to scalse Kafka clsuter without upgrade, that increses time of scale.
In production this can become critical and lead to downtime. For example last 2 times the scale lasted for 12 hours.1 vote -
Kafka tiered storage with external S3 bucket
As a developer
I want to have the ability to use our own S3 bucket for storing Kafka tiered storage
so that I can access the data from S3 and query some data for debugging (without streaming all the data to Kafka)3 votes -
kafka_connect_connector_metrics availability over Prometheus
Our customer Jago wants to monitor the status of connectors and tasks but currently can't find relevant metrics to do so. They want to be able to monitor the status of connectors and tasks on a dashboard and also get notified whenever a connector has not been running for X minutes.
Jago has a connector running but can not find the metrics for kafka.connect:type=connector-metrics,connector=*.
The specific metrics they are looking for is the one related to the status of a connector. For example, in the customers current self-managed kafka connect, they have the following metrics. This is convenient because they…
1 vote -
Include custom Kafka topic tags as Prometheus metric labels
The Aiven platform allows custom key/value tags to be added to resources such as Kafka topics. It would be useful to have these exposed as additional labels on metrics so that alerts can be triggered based on this metadata.
3 votes -
Kafka rack awareness
Kafka consumer clients on Aiven for Kafka should be able to use client.rack configuration introduced in KIP-392
3 votesShelved ·not currently possible in Kafka natively, will re-evaluate if and when becomes available
-
Kafka rate limits and quotas configuration via Terraform
As a Cloud platform engineer I need to provide the capability, to developers, to set quotas via Terraform when then set-up applications that will produce to/consume from a Kafka cluster.
3 votes -
Automated dynamic quota configuration
As a Cloud platform engineer, I need to have an automated way to set up and update quota configurations on a cluster, taking into account changes in resources consumption patterns amongst producers and consumers.
1 vote -
Visualization of client IDs that interact with a Kafka cluster
As a Cloud platform engineer, I need to have a view of Kafka cluster resources consumption across all clients IDs. Such information is needed to configure quotas and to manage them.
1 vote -
Configure CPU threshold for alert mails in kafka
As a kafka user, I want to configure the threshold used to send CPU warning emails. Right now we are getting emails because the CPU exceeds 50%, which is irrelevant. We know we can configure the emails to go to another mail address but in that case we would loose all other technical mails as well.
It would also be good for us if we can turn off that single alert.2 votes
- Don't see your idea?