tencent cloud

TDMQ for CKafka

Release Notes and Announcements
Release Notes
Broker Release Notes
Announcement
Product Introduction
Introduction and Selection of the TDMQ Product Series
What Is TDMQ for CKafka
Strengths
Scenarios
Technology Architecture
Product Series Introduction
Apache Kafka Version Support Description
Comparison with Apache Kafka
High Availability
Use Limits
Regions and AZs
Related Cloud Services
Billing
Billing Overview
Pricing
Billing Example
Changing from Postpaid by Hour to Monthly Subscription
Renewal
Viewing Consumption Details
Overdue Payments
Refund
Getting Started
Guide for Getting Started
Preparations
VPC Network Access
Public Domain Name Access
User Guide
Usage Process Guide
Configuring Account Permission
Creating Instance
Configuring Topic
Connecting Instance
Managing Messages
Managing Consumer Group
Managing Instance
Changing Instance Specification
Configuring Traffic Throttling
Configuring Elastic Scaling Policy
Configuring Advanced Features
Viewing Monitoring Data and Configuring Alarm Rules
Synchronizing Data Using CKafka Connector
Use Cases
Cluster Resource Assessment
Client Practical Tutorial
Log Integration
Open-Source Ecosystem Integration
Replacing Supporting Route (Old)
Migration Guide
Migration Solution Overview
Migrating Cluster Using Open-Source Tool
Troubleshooting
Topics
Clients
Messages
​​API Reference
History
Introduction
API Category
Making API Requests
Other APIs
ACL APIs
Instance APIs
Routing APIs
DataHub APIs
Topic APIs
Data Types
Error Codes
SDK Reference
SDK Overview
Java SDK
Python SDK
Go SDK
PHP SDK
C++ SDK
Node.js SDK
SDK for Connector
Security and Compliance
Permission Management
Network Security
Deletion Protection
Event Record
CloudAudit
FAQs
Instances
Topics
Consumer Groups
Client-Related
Network-Related
Monitoring
Messages
Agreements
CKafka Service Level Agreements
Contact Us
Glossary

Slow Consumption Speed

PDF
Focus Mode
Font Size
Last updated: 2026-01-20 17:19:22

Overview

The consumer consumes messages slowly.

Possible Causes

High server load
Traffic throttling issue
Client load
Consumer processing capacity issue
Network issue

Solutions

High server load
To confirm whether the cause is a server issue, you can view the consumption time in Advanced Monitoring in the console. The time information indicates the time taken by the server to process requests. If there is an issue with the server load, you can observe higher time statistics at each processing stage, as shown in the figure below:

Traffic throttling issue To confirm whether the cause is a traffic throttling issue, you can configure a bandwidth over-limit alarm. Choose Monitoring > Instance to check whether the peak bandwidth of the instance has been reached. If the peak bandwidth has been reached, you need to upgrade the instance's peak bandwidth. For how to upgrade instance configurations, see Changing Instance Specifications.

Client load
If there is no performance issue with the server, the cause is probably the insufficient consumption capacity of the client. First, check the mapping between partitions and consumers. If a single consumer is consuming too many partitions, it is recommended to increase the number of consumers. Try to make one consumer consume only one partition. View the partition-to-consumer mapping, as shown in the figure below:

Consumer processing capacity issue If the partition-to-consumer assignment is normal, you can scale out partitions in the console to increase data consumption parallelism. Partition scale-out in the console is instantaneous and lossless, with no impact on your business. Scale out partitions as shown in the figure below:

Network issue
Check the client load. For example, check metrics such as the CPU, memory, and network interface card (NIC) on the client machine. For a Java process, focus on the usage of GC and heap memory.
Consumption linkage issue
Check the downstream segment of the message consumption linkage for other causes for the slow consumption speed. For example, consuming messages requires certain synchronous operations to be completed before submitting the offset to the server. Common abnormal downstream operations include:
Logs are printed frequently.
Database writes are slow.
The business logic is complex, causing slow message processing.
If the issue cannot be resolved quickly, it is recommended to increase the number of topic partitions and scale out consumer instances to rapidly boost the message consumption speed.

Help and Support

Was this page helpful?

Help us improve! Rate your documentation experience in 5 mins.

Feedback