[STORM-3649] Logic error regarding storm.supervisor.medium.memory.grace.period.ms

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.0.0, 2.1.0, 2.2.0
• Fix Version/s: 2.3.0
• Component/s: storm-server
• Assignee: Ethan Li
• Reporter: Ethan Li
• Created: 08/Jun/20 20:46
• Updated: 01/Jul/20 13:57
• Resolved: 09/Jun/20 20:27

Description

Inside this chunk of code

https://github.com/apache/storm/blob/2.2.x-branch/storm-server/src/main/java/org/apache/storm/daemon/supervisor/BasicContainer.java#L758

if (systemFreeMemoryMb < mediumMemoryThresholdMb) {
                    if (memoryLimitExceededStart < 0) {
                        memoryLimitExceededStart = Time.currentTimeMillis();
                    } else {
                        long timeInViolation = Time.currentTimeMillis() - memoryLimitExceededStart;
                        if (timeInViolation > mediumMemoryGracePeriodMs) {
                            LOG.warn(
                                "{} is using {} MB > memory limit {} MB for {} seconds",
                                typeOfCheck,
                                usageMb,
                                memoryLimitMb,
                                timeInViolation / 1000);
                            return true;
                        }
                    }
                } 

At very beginning, memoryLimitExceededStart in BasicContainer is initialized as 0. :
https://github.com/apache/storm/blob/2.2.x-branch/storm-server/src/main/java/org/apache/storm/daemon/supervisor/BasicContainer.java#L80

protected volatile long memoryLimitExceededStart;

So once it hits this scenario, the grace period doesn't really take any effect because the timeInViolation will be very large (equals to currentTime)

The logs from a test:

2020-06-08 20:39:18.277 o.a.s.d.s.BasicContainer SLOT_6707 [WARN] WORKER 9c16e81e-4936-4029-bcda-ceb5b74b8f42 is using 167 MB > memory limit 158 MB for 1591648758 seconds