Merge ~rodsmith/plainbox-provider-checkbox:update-disk-read-performance-thresholds into plainbox-provider-checkbox:master
Status: | Merged |
---|---|
Approved by: | Rod Smith |
Approved revision: | 9179cb98372f454eeccba983d29da45bb615da01 |
Merged at revision: | f71686e8e4f137e2380459d37f0f808adb43013a |
Proposed branch: | ~rodsmith/plainbox-provider-checkbox:update-disk-read-performance-thresholds |
Merge into: | plainbox-provider-checkbox:master |
Diff against target: |
40 lines (+14/-0) 1 file modified
bin/disk_read_performance_test (+14/-0) |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Rod Smith | Approve | ||
Jonathan Cave (community) | Approve | ||
Jeff Lane | Approve | ||
Review via email: mp+381120@code.launchpad.net |
Commit message
Update disk_read_
Description of the change
Added device types for NVDIMM, SSD, ATA, & SCSI devices -- the latter two were identified by the script but fell back to the default (15 MB/s) pass criterion, and the first two require new code to identify.
This MR also sets new or updates pass criteria for NVMe, NVDIMM, MDADM, ATA, SCSI, and SSD devices, based on data from over 50 devices in a semi-random sample -- see:
https:/
Criteria were set a little lower than the lowest observed value for most devices, with an exception being HDD (ATA and SCSI) devices, which were set above the pathetic level of a couple of very old disks and a bit below the lowest observed value for more modern hardware. I set the NVDIMM value to match the NVMe value because I have access to just one machine with NVDIMM hardware, and average performance levels seemed similar.
LGTM. I'd suggest at least letting Devices know this changed, but it'll sit in Dev for a while and they can test it and change if necessary.