Method for batched server memory fault detection through IPMI tool

Abstract

The invention provides a method for batched server memory fault detection through an IPMI tool, and belongs to the field of fault detection. The method comprises the steps that recording and scanning bmc logs of all servers inside a network through the IPMI tool, analyzing machines with the memory problem from the result, conducting batched checking on the batched servers inside the network through a script, rapidly determining the machines with the memory ecc errors, and achieving memory batched checking of the batched machines. According to the method, the testing time is shortened, and working efficiency is improved.

Description

technical field [0001] The present invention relates to a method for batch memory problem detection under the condition that a batch deployment server has a bmc record memory fault function, specifically a method for using an IPMI tool to detect a batch server memory fault. Background technique [0002] In computers, Machine Check Architecture (MCA) refers to a mechanism for the CPU to report hardware errors in the operating system. It is a ras feature of the CPU; for example, when an ECC error occurs, such as a memory error, it is located in the CPU. Various model-specific registers (MSRs) will detect errors and trigger the MCA mechanism; then a system interrupt will be generated, and various state information at that time will be recorded by various registers (MSRs) and handed over to the bmc chip for recording , so the current motherboard integrated bmc chip can record memory operation errors, especially ECC error reporting, bmc has an independent network configuration,...

AI Technical Summary

Problems solved by technology

[0003]At present, a large number of Internet users purchase batch servers, and with the gradual maturity of remote management technology, the management of servers no longer depends on the local management of the computer room where the server is located, but through Network remote control, so that when the server has a memory failure such as ECC ERROR error, if the bmc check is not passed, the problem cannot be found in time, which may affect the stability of the later server operation, so it is necessary to regularly check the bmc log of all servers, But for machines deployed in batches, it takes too long to test one by one, and the work efficiency is too low