Merge lp:~sylvain-pineau/checkbox/add_removable_attribute_to_udev_resource into lp:checkbox
Status: | Rejected |
---|---|
Rejected by: | Sylvain Pineau |
Proposed branch: | lp:~sylvain-pineau/checkbox/add_removable_attribute_to_udev_resource |
Merge into: | lp:checkbox |
Diff against target: |
108 lines (+22/-7) 4 files modified
checkbox/parsers/udevadm.py (+12/-0) jobs/disk.txt.in (+7/-5) scripts/disk_read_performance_test (+1/-1) scripts/udev_resource (+2/-1) |
To merge this branch: | bzr merge lp:~sylvain-pineau/checkbox/add_removable_attribute_to_udev_resource |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Marc Tardif (community) | Needs Information | ||
Review via email:
|
Description of the change
After reviewing several systems which reported disk_read_
An easy fix is to lower this value to 7 MB/s.
Another proposal is to run disks.txt.in jobs only on non-removable devices, this is the purpose of the changes to udev_resource and its corresponding parser, add a new attribute we could use in job requirements.
I'm still not sure about what to do for removable devices, duplicating the read_performance test in their related sections (usb, firewire, etc...), add benchmarks-like tests to removable_
So several solutions for one problem, that's why i'm requesting your reviews.
Thanks.
Unmerged revisions
- 1396. By Sylvain Pineau
-
Add the removable attribute to udev_resource and a new requirement to all jobs in disk.txt.in to only run on non-removable devices.
I don't think we should do benchmarking on removable devices (yet; we may have to when we start testing USB 3.0). The reason is that this is very dependent on the device itself, I've seen USB sticks with transfer rates as low as 1 MB/s, especially cheap ones. On the other hand, performance for internal hard drives is pretty much standard so we *can* benchmark those more reliably.