100% CPU usage, JVM error 'OutOfMemoryError : GC overhead limit exceeded'

Hello everybody,

since version nexus-3.23.0-03, currently we are using nexus-3.24.0-02, nexus get stuck frequently and i do not see the cause.

We are running a CentOS 7.7 machine with openjdk version “1.8.0_242”, 16Cores, 32Gb Ram, JVM settings as provided in your documentation, installed with Distribution Archive, has 4 blobstores (500GB, 3x100GB) and also act as a proxy.

Nexus become unresponsive, all CPU Cores running at 100%, RAM usage is about 16Gb.
In logfile are many “GC overhead limit exceeded” messages regarding database. Checking database health does not show any errors.

2020-06-26 09:54:21,933+0200 ERROR [qtp54244712-2064] *UNKNOWN org.sonatype.nexus.repository.npm.internal.NpmAuditErrorHandler - java.util.concurrent.ExecutionException: java.lang.OutOfMemoryError: GC overhead limit exceeded
java.lang.RuntimeException: java.util.concurrent.ExecutionException: java.lang.OutOfMemoryError: GC overhead limit exceeded

2020-06-26 10:07:54,789+0200 ERROR [qtp54244712-2234] *UNKNOWN org.sonatype.nexus.repository.storage.LastDownloadedHandler - Failed to update last downloaded time for request /rimraf/-/rimraf-3.0.2.tgz
com.orientechnologies.orient.core.exception.OJVMErrorException: JVM error ‘OutOfMemoryError : GC overhead limit exceeded’ occurred during data processing, storage is switched to ‘read-only’ mode. To prevent this exception please restart the JVM and check data consistency by calling of ‘check database’ command from database console.
DB name=“component”

what is your preferred way to upload the log file?

do you have any ideas or hints where to look at?

thx in advance
marco

You can generate a support zip that contains all logs in administration menu and attach it here (it will be publicly visible), on our issue tracker or preferably using official support channel if you’re a PRO customer.

thanks for your reply.
the upload feature does not allow zip files, but i will create an issue in tracker.

@marco.huebl Did you create an issue? I think we’re seeing something similar.