Agent with Windows virtual node crashes when streaming to a parent

I have a Windows VM streaming to a Netdata Agent successfully with the latest Prometheus windows exporter. Metrics for both machines appear in the local dashboard without any issues. I want to make this agent a child of another agent; however, when I configure streaming, the agent crashes on the child. The parent records enough about the child to display it in the dashboard, but it displays as offline, and no metrics are received. The local web service on the child also fails to load after streaming is enabled.

If I remove the Windows virtual node from the child’s configuration, streaming works as expected and the parent displays metrics from the child.

I’d like both the child and the virtual nodes it pulls data from to roll up into the dashboard on the parent. Is this configuration supported?

The error log cycles through the messages below:

2023-06-12 14:17:53: netdata ERROR : PLUGINSD : cannot open plugins directory '/etc/netdata/custom-plugins.d' (errno 2, No such file or directory)
2023-06-12 14:17:53: netdata ERROR : PD[perf] : child pid 640111 exited with code 1.
2023-06-12 14:17:53: netdata ERROR : PD[perf] : PLUGINSD: 'host:netdata-t03', '/usr/libexec/netdata/plugins.d/perf.plugin' (pid 640111) exited with error code 1 and haven't collected any data. Disabling it.
2023-06-12 14:17:53: netdata ERROR : PD[ioping] : child pid 640115 exited with code 1.
2023-06-12 14:17:53: netdata ERROR : PD[ioping] : PLUGINSD: 'host:netdata-t03', '/usr/libexec/netdata/plugins.d/ioping.plugin' (pid 640115) exited with error code 1 and haven't collected any data. Disabling it.
2023-06-12 14:17:54: netdata ERROR : STATS_GLOBAL : STREAM netdata-t03 [send]: not ready - collected metrics are not sent to parent.
2023-06-12 14:17:55: netdata ERROR : PD[go.d] : STREAM sql01 [send]: not ready - collected metrics are not sent to parent.

And the output from the service status a few seconds after the service starts:

netdata.service - Real time performance monitoring
     Loaded: loaded (/usr/lib/systemd/system/netdata.service; enabled; preset: enabled)
     Active: activating (auto-restart) (Result: core-dump) since Mon 2023-06-12 14:06:36 PDT; 12s ago
    Process: 632146 ExecStartPre=/bin/mkdir -p /var/cache/netdata (code=exited, status=0/SUCCESS)
    Process: 632147 ExecStartPre=/bin/chown -R netdata /var/cache/netdata (code=exited, status=0/SUCCESS)
    Process: 632148 ExecStartPre=/bin/mkdir -p /run/netdata (code=exited, status=0/SUCCESS)
    Process: 632149 ExecStartPre=/bin/chown -R netdata /run/netdata (code=exited, status=0/SUCCESS)
    **Process: 632150 ExecStart=/usr/sbin/netdata -P /run/netdata/netdata.pid -D (code=dumped, signal=ABRT)
   Main PID: 632150 (code=dumped, signal=ABRT)**
        CPU: 2.258s

From /var/log/messages:

Jun 12 14:26:13 netdata-t03 systemd-coredump[646015]: Resource limits disable core dumping for process 645654 (netdata).
Jun 12 14:26:13 netdata-t03 systemd-coredump[646015]: Process 645654 (netdata) of user 978 dumped core.

Agent versions:
Parent: netdata v1.39.1
Child (crashing): netdata v1.39.0-116-nightly

Big thanks for reporting, @backchapman. I opened an issue on GH.

Fixed in 15183.

Confirmed it is working as expected now. Data from the child and its virtual node are rolling up into the parent. Thank you @ilyam8 for the quick fix!

I’m experiencing this error message with an Nginx stream/upstream/server setup in the central node. None child nodes can report to the parent no more.