Description of problem: Customer is noting on their RHEL 8.2 laptop with 16 GiB of memory and no swapspace that the system hangs for an indeterminate amount of time. Through investigations, it appears the particular workload of the customer within firefox is causing firefox to consume memory gradually until the point of oom killer invocation. That being said, the time between the beginning of doing direct reclaim and forcing pretty severe major faulting activity systemwide up to the oom killer invocation can be quite variant. Version-Release number of selected component (if applicable): 4.18.0-193.1.2.el8_2.x86_64 How reproducible: Pretty reliable Steps to Reproduce: 1. Have the customer run firefox with a large number of tabs and cause pretty severe memory pressure 2. wait until system begins to hang 3. Actual results: System can hang for a few minutes to over an hour Expected results: customer expects the system to recover quickly Additional info: Customer has been provided a workaround with using cgroups to limit the memory usage of firefox. However, the hangs can still happen and firefox is killed more often.
Hey all, Closing out the BZ due to lack of response from the customer. Thank you all for the help internally on this matter.