Merge ~harlowja/cloud-init:sys-io-errors into cloud-init:master
| Status: | Merged |
|---|---|
| Merged at revision: | 3ac34f5325edf7f1212cd0b6a3d1cfe2ed45a63b |
| Proposed branch: | ~harlowja/cloud-init:sys-io-errors |
| Merge into: | cloud-init:master |
| Diff against target: |
333 lines (+102/-95) 3 files modified
cloudinit/net/__init__.py (+84/-82) cloudinit/net/cmdline.py (+6/-3) tests/unittests/test_net.py (+12/-10) |
| Related bugs: |
| Reviewer | Review Type | Date Requested | Status |
|---|---|---|---|
| Lars Kellogg-Stedman (community) | Approve on 2016-12-02 | ||
| Scott Moser | 2016-09-15 | Needs Fixing on 2016-11-02 | |
|
Review via email:
|
|||
| Scott Moser (smoser) wrote : | # |
| Joshua Harlow (harlowja) wrote : | # |
Ya, that makes sense to me, I can adjust this one.
| Scott Moser (smoser) wrote : | # |
hm.. i had just inteded to say lets drop sys_netdev_info and instead make the callers use read_sys_net
| Joshua Harlow (harlowja) wrote : | # |
Oh, lol.
| Mike Dorman (mdorman-m) wrote : | # |
Data point: We tried this patch out locally under RHEL7 and it's working like a champ.
| Scott Moser (smoser) wrote : | # |
it looks like your test case is busted, flake8 fails which makes me wonder if its working correctly.
tests/
tests/
that and a few other changes:
http://
we should not any longer need to catch the IOError and OSError from users of read_sys_net as it should handle them if enoent is provided.
| Scott Moser (smoser) wrote : | # |
Josh,
can you take another look here?
| Joshua Harlow (harlowja) wrote : | # |
Will double check, though I fixed that recently.
| Joshua Harlow (harlowja) wrote : | # |
*thought
| Scott Moser (smoser) wrote : | # |
it looks like you fixed the test cases, but one error still.
Also, can you adjust summary line of your commit to be
Replace usage of sys_netdev_info with read_sys_net.
| Lars Kellogg-Stedman (larsks) wrote : | # |
The updated patch looks good to me. One inline comment, but it's not operationally important.


Josh,
I had sort of done this as: paste.ubuntu. com/23207571/
http://
The person who had reproduced this on arch said that the patch above it fixes it.
Another option...
it looks like callers of sys_netdev_info could use read_sys_net.
And read_sys_net actually handles the IOError already.
I think thats the option I'd prefer.