Container Failure Cause Identification via Process Symbol Aggregation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In cloud computing systems, existing methods struggle to distinguish between application process failures within containers and infrastructure issues, as they cannot accurately differentiate between processes in the container and the host OS, making it difficult to identify the root cause of abnormalities.
Innovation Solution
A failure cause identification program that collects process information, generates symbol information associating functions with containers, aggregates frequency data, and identifies failure causes by analyzing this data to determine if issues lie within the application or the infrastructure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing monitoring methods are used in container environments, then general process information can be collected, but the system cannot distinguish between container processes and host OS processes, making root cause identification difficult
Solution Approach 1:
The patent segments the process information by creating separate data structures for container processes and host OS processes. It uses process ID (PID) namespace isolation to divide the monitoring scope, allowing independent tracking of container processes within their own namespace while distinguishing them from host processes. This segmentation enables precise identification of whether a failure originated from container or host infrastructure.
Solution Approach 2:
The patent introduces an intermediary mechanism - the PID namespace mapping table - that acts as a mediator between the container process space and host process space. This intermediary structure maintains the correspondence between container PIDs and host PIDs, enabling accurate attribution of process actions to their correct origin without direct confusion between the two process spaces.
2Measurement precision
If detailed process information is collected for each container, then failure cause identification improves, but the data aggregation and analysis complexity increases
Solution Approach 1:
The patent merges the monitoring of multiple container processes and host processes into a unified data structure that preserves individual process information while enabling collective analysis. The process information collection module aggregates data from various sources (container runtime, host OS) into a standardized format, and the analysis module combines this information with namespace context to provide comprehensive failure diagnosis without manual intervention.
Solution Approach 2:
The patent creates a virtual copy of the process namespace hierarchy through data structures that mirror the actual container and host process relationships. Instead of directly accessing complex native process information, the system maintains simplified copies of process metadata, PID mappings, and namespace relationships that can be analyzed without interfering with the actual running processes.
3Ease of operation
If process information is collected without container context, then data collection is simple, but the system cannot determine whether failures are application-specific or infrastructure-wide
Solution Approach 1:
The patent adds a new dimension to process information by incorporating namespace context. Instead of only collecting traditional process metrics, the system enriches each process record with namespace identification (container ID, host ID) and hierarchical relationship data. This additional dimensional information allows the system to determine failure scope - whether an issue is isolated to a specific container application or affects the host infrastructure - without complicating the basic data collection mechanism.
Data Source
AI summary
A non-transitory computer-readable recording medium stores a failure cause identification program for causing a computer to execute a process including: collecting process information related to one or more processes that operate in a container environment; obtaining a derivative relationship of a process for each container on the basis of the process information; generating symbol information in which a function of each of the processes is associated with a container in which each of the processes operates according to the derivative relationship of the process for each container; generating an aggregation result in which a frequency of the function is aggregated according to the symbol information; and identifying a cause at a time of failure occurrence on the basis of the aggregation result.