Closed wilzbach closed 6 years ago
It's hard to be sure on associating kernel logs to test cases, but I do see some oom kills of dmd, so possibly. The host in question there has 4 gigs of memory, which is more than some of the other linux build hosts and less than others. But if that run failed, then something changed because it's not a new host or new host type... been using the same ec2 instance size for years.
Generally speaking, you guys need to get your memory consumption under control.
Need a better allocator than the current bump the pointer and never-free strategy.
resolving this issue.. it's not a tester problem but rather the system under test.
Looks like it's running out of memory?
https://auto-tester.puremagic.com/show-run.ghtml?projectid=1&runid=2998102&isPull=true