Bug 1366298
Summary: | Prevent creation of a bond with vlan slaves | ||
---|---|---|---|
Product: | Red Hat Enterprise Linux 7 | Reporter: | Fabian Deutsch <fdeutsch> |
Component: | NetworkManager | Assignee: | Beniamino Galvani <bgalvani> |
Status: | CLOSED NOTABUG | QA Contact: | Desktop QE <desktop-qa-list> |
Severity: | high | Docs Contact: | |
Priority: | high | ||
Version: | 7.2 | CC: | atragler, bgalvani, danken, fdeutsch, huzhao, lrintel, rkhan, sukulkar, thaller, ycui |
Target Milestone: | rc | ||
Target Release: | --- | ||
Hardware: | Unspecified | ||
OS: | Unspecified | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | If docs needed, set a value | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2016-10-06 09:36:58 UTC | Type: | Bug |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: |
Description
Fabian Deutsch
2016-08-11 14:36:24 UTC
We've seen this a couple of times in RHEV, up to now we prevented this in high-level tools, but it should actually be prevented by NetworkManager to create unsupported/non-working setups. I do see there are combinations that do work - so we might not want to prevent this setp completely. OTOH there are combinations which don't work, and we should prevent that a user is setting them up. As said, we do see users trying this in the real world, and they do run into issues. We could block it on the high-level, but I'd rather like to see it fixed at the root - so either NM/legacy initscripts or even in kernel land. (In reply to Fabian Deutsch from comment #3) > I do see there are combinations that do work - so we might not want to > prevent this setp completely. > > OTOH there are combinations which don't work, and we should prevent that a > user is setting them up. > > As said, we do see users trying this in the real world, and they do run into > issues. > > We could block it on the high-level, but I'd rather like to see it fixed at > the root - so either NM/legacy initscripts or even in kernel land. After a bit of research, I don't think there aren't any technical reasons why a bond over VLANs should not work. We don't support this configuration in RHEL [1], and instead favor the use of VLAN over bond. Nevertheless the setup is legitimate and NetworkManager should not prevent it. Do you have more details about which configurations have caused issues? [1] https://bugzilla.redhat.com/show_bug.cgi?id=952719 Well, in bug 1356635 we see that a bond over VLANs leads to issues. In our experience this setup is _never_ stable (having a bond over vlan devices [slaves]). Dan or Ying might recall more bugs. I'm afraid I do not recall. But unrelated to this, oVirt does not support (nor even recognizes) bond-over-vlan, and I'm not aware of anybody asking us to add this. For our use case, it would be cool if Cockpit had a way to block it. Okay, so let's see if Ying has a good example for failures. Otherwise it seems that we don't agree to prevent this setups, and thus we can then close this bug. Huijuan, See this bug, could you provide our thoughts for comment 5 and comment 7 based on our testing? (In reply to Fabian Deutsch from comment #5) > Well, in bug 1356635 we see that a bond over VLANs leads to issues. > > In our experience this setup is _never_ stable (having a bond over vlan > devices [slaves]). > > Dan or Ying might recall more bugs. In my test, can not setup bond over VLANs via Cockpit, in other word, it is blocked by Cockpit. Blocking the creation of unwanted configurations from upper layers, as cockpit does, makes more sense to me. I'm closing this for now, please reopen if there are any updates or more specific requests. |