Strange Full GC and OutOfMemory issue

Aleksey Shipilev shade at redhat.com
Tue Apr 10 15:38:12 UTC 2018


Hi Omar,

On 04/05/2018 12:49 AM, Omar Kilani wrote:
> We're (Remember The Milk) using JDK9/10 + Shenandoah for all of our
> services. Thanks for making this available -- it's amazing! :)

Cool, any performance data you can share?

> We've recently started running into a weird issue where Shenandoah
> thinks it needs to do a Full GC after running for a a week or so. I
> think it's when there's a burst of memory allocation, but I'm not 100%
> sure.

Yes, that is possible, it is usually visible in GC logs.

Until recently (~3 wks ago), the allocation spike could trip Shenandoah into Full GC. It is
hopefully better now. You might want to read this section on our Wiki:
  https://wiki.openjdk.java.net/display/shenandoah/Main#Main-FailureModes

Please try to upgrade and see if the issue still persists.


> I'm attaching a gc.log and a safepoint.log in case it helps. Please
> let me know what else we can do to provide more details.

Our OpenJDK mailing lists scrub big/suspicious attachments, alas. Is there a chance you can put the
log somewhere on the web and put the link here?

Thanks,
-Aleksey


More information about the shenandoah-dev mailing list