Stopping Elasticsearch causes Windows Server to crash/restart

[mbalzary]

Hi,

we installed Elasticsearch 7.10 on a Windows Server 2016. Only configuration changes made are regarding cluster.name and node.name. Elasticsearch can be started without a problem and runs, but when it is stopped it causes the whole server to crash and reboot. This happens when starting/stopping it via command line and when using a service.

The final log lines of Elasticsearch before the crash are:

[2021-04-22T13:24:00,910][INFO ][o.e.g.GatewayService     ] [node-2] recovered [8] indices into cluster_state
[2021-04-22T13:24:04,118][INFO ][o.e.c.r.a.AllocationService] [node-2] Cluster health status changed from [RED] to [YELLOW] (reason: [shards started [[.kibana-event-log-7.10.0-000001][0], [.kibana_1][0]]]).
[2021-04-22T13:24:15,287][INFO ][o.e.c.m.MetadataIndexTemplateService] [node-2] adding template [.management-beats] for index patterns [.management-beats]

After that nothing more is logged. On another Windows Server 2016 this behavior does not occur and its Elasticsearch logs look equal to the ones of the crashed server.

The crash happens with and without having Kibana running. Changing the JVM Heap in the jvm.options file did not change the behavior. Trying it with Elasticsearch 7.11 also led to the crash/reboot of the machine. In the Windows Event Viewer there were no helpful log information.

Has someone seen this before or has an idea of what could cause the issue?

Thank you and regards

[warkolm (Mark Walkom)]

Welcome to our community!

Is there anything in your OS logs that might give an indication?

[mbalzary]

Thank you

Unfortunately, there is nothing in the OS logs to give an indication.

[DavidTurner]

There is something fundamentally wrong with your machine: there's no way that a user-space process like Elasticsearch can cause a healthy machine to crash. I would suspect faulty hardware (maybe CPU or RAM) but in any case this doesn't sound like something we can fix on the Elasticsearch side.