A customer recently submitted a case that Pedro Principeza, a Canonical Support Engineer, was able to solve using an interesting technique. Principeza was contacted to help pinpoint the processes that were generating a high rate of I/O operations, affecting server performance.
Implementing a trace to pinpoint workload usage was the best solution, however, resources on its set up were minimal. In order to formulate a thorough manner of tracking activity tied to two specific parts of the kernel, Principeza researched and dived into the Linux kernel official documentation.
Principeza had experience with tracing capabilities and knew the next step would be simulating the problem locally on Linux. Once the behavior was simulated in Canonical’s lab environment, he was able to deliver the kernel tracing procedure in 30 minutes.
After parsing through a lot of documentation, recreating the problem with Linux and tapping into previous experience utilising tracing capabilities, Principeza solved the case.
The trace revealed that two applications, one security and one SDN-type application, were draining server resources. Upon detecting which applications were causing the high I/O rates, he engaged the applications’ support (outside Canonical) in order to fully understand the causes of the erratic behaviour to provide recommendations on reducing server consumption going forward.
Tracking this type of resource usage requires a medium to high-level of expertise, so having additional engineering resources via technical support from experts like Principeza is a major value add to businesses working through complex IT issues.
To provide guidance on similar cases in the future, the following articles are available to customers in our Knowledge Base*: