[Bug] BE IO metrics are missing when storing data on nvme devices #48350
Description
Search before asking
- I had searched in the issues and found no similar issues.
Version
2.1.7
What's Wrong?
When deploying Doris BE on machines with NVME disks, I/O metrics like doris_be_max_disk_io_util_percent or doris_be_disk_io_time_ms are all 0's.
After debugging, it is found that device labels in the metrics are like nvmeXnYp.
$ curl http://localhost:8040/metrics | grep "doris_be_disk_io_time_ms"
# TYPE doris_be_disk_io_time_ms counter
doris_be_disk_io_time_ms{device="nvme1n1p"} 0
doris_be_disk_io_time_ms{device="nvme5n1p"} 0
doris_be_disk_io_time_ms{device="nvme0n1p"} 0
doris_be_disk_io_time_ms{device="nvme3n1p"} 0
doris_be_disk_io_time_ms{device="nvme6n1p"} 0
doris_be_disk_io_time_ms{device="nvme7n1p"} 0
doris_be_disk_io_time_ms{device="nvme4n1p"} 0
doris_be_disk_io_time_ms{device="nvme2n1p"} 0
while the real device names are like nvmeXnY or nvmeXnYpZ.
$ cat /proc/partitions
major minor #blocks name
8 0 468320256 sda
8 1 488264 sda1
8 2 467830784 sda2
253 0 20971520 dm-0
253 1 12582912 dm-1
253 2 434274304 dm-2
259 0 1875374424 nvme0n1
259 1 1875373056 nvme0n1p1
259 2 1875374424 nvme1n1
259 3 1875373056 nvme1n1p1
259 4 1875374424 nvme7n1
259 5 1875373056 nvme7n1p1
259 6 1875374424 nvme6n1
259 7 1875373056 nvme6n1p1
259 8 1875374424 nvme5n1
259 12 1875373056 nvme5n1p1
259 9 1875374424 nvme4n1
259 13 1875373056 nvme4n1p1
259 10 1875374424 nvme3n1
259 14 1875373056 nvme3n1p1
259 11 1875374424 nvme2n1
259 15 1875373056 nvme2n1p1
What You Expected?
expose I/O metrics correctly for NVME devices
How to Reproduce?
No response
Anything Else?
No response
Are you willing to submit PR?
- Yes I am willing to submit a PR!
Code of Conduct
- I agree to follow this project's Code of Conduct