[Bug] Broker load rebalance uneffective in virtual machine，failed to find NICs while "/virtual/" in the realpath
- Dominant language
- Java
- Stars
- 15.3k
- Forks
- 3.8k
- Avg merge
- 1d 14h
- Merged PRs (30d)
- 160
Description
### Search before asking
- [X] I searched in the [issues](https://github.com/apache/pulsar/issues) and found nothing similar.
### Version
tlinux(CentOS)
pulsar-2.10.1
### Minimal reproduce step
Deploy brokers on virtual machines.
### What did you expect to see?
Broker auto load rebalance effective.
### What did you see instead?
According to the code, seems that we can't get bandwidthin&out correctlly in virtual machine , which made the broker load rebalance uneffective with ThresholdShedder strategy.
### Anything else?
It's there any plans to support this scene？Or some methods to make it effective in virtual machine?
### Are you willing to submit a PR?
- [ ] I'm willing to submit a PR!
Contributor guide
Research direction
Start by reproducing the issue with brokers deployed on virtual machines using Pulsar 2.10.1 on CentOS, then inspect the broker load-rebalance path and the ThresholdShedder strategy mentioned in the report. Compare bandwidth-in and bandwidth-out detection with the reported realpath containing "/virtual/". Done means identifying why NICs are not found and confirming effective broker load rebalancing in this environment.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- java
- Domain
- distributed-systems
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 30/100