Package "/sys/bus/node/devices/node0" CPU "0" is missing package_throttle / panic: runtime error: index out of range #745
Comments
|
It looks like this is a problem for the thermal throttle part of the CPU collector. This part of the code was refactored recently. Would it be possible for you to test a build from latest master? It would also be useful to post the output of |
|
Where can I find the latest master build (binary)? Output from a working node: Output from a not working node: |
|
For master, you will have to build your own. Instructions are in the README. |
|
@stephan2012 Please show us the output of this command from both systems: for i in {0..1}; do echo -n "$i: " ; cat /sys/bus/node/devices/node$i/cpulist ; doneTake a look at PR #734 which will probably fix the crash as your non-working system has no cpus in “node1”. Your working system has no 2nd node “node1” (only a single “node0”). |
|
Output from working node: # for i in {0..1}; do echo -n "$i: " ; cat /sys/bus/node/devices/node$i/cpulist ; done
0: 0-15
1: cat: /sys/bus/node/devices/node1/cpulist: No such file or directoryOutput from not working node: # for i in {0..1}; do echo -n "$i: " ; cat /sys/bus/node/devices/node$i/cpulist ; done
0: 0-15
1: So PR #734 could solve my issue but no chance to build latest master so far. Trying to do this today. |
|
@stephan2012 Yes, an empty You may also want to talk to your hosting provider because this is IMHO a strange system config. Make sure this is intentional on their side. |
|
Latest build from master works. No more panics. Thanks, guys. |
|
Great, hopefully we can make an official release with this fix soon. Since it's technically a breaking change, it will need to be in a minor release. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Host operating system: output of
uname -anode_exporter version: output of
node_exporter --versionnode_exporter command line flags
tried with
--no-collector.hwmonafter reviewing #697.Are you running node_exporter in Docker?
No.
What did you do that produced an error?
Run node_exporter as shown above. Wait from Prometheus to scrape metrics or curl/wget to the metrics endpoint.
What did you expect to see?
node_exporter running without panic. :-)
What did you see instead?
Panic:
What is quite strange: We have 10 identical systems hosted on VMware and running the system OS (RHEL 7.4) and patch level. The 4 older systems run node_exporter without issues, the new 6 systems face the issue described above. Not sure if there is a difference in the VMware hardware version because these VMs are provided by our hosting provider.
The text was updated successfully, but these errors were encountered: