Details
- Reviewers
kevans
Diff Detail
- Repository
- rS FreeBSD src repository - subversion
- Lint
Lint Passed - Unit
No Test Coverage - Build Status
Buildable 37773 Build 34662: arc lint + arc unit
Event Timeline
I saw the upstream commit to use the ifnet creator's creds for the socket, so one can keep the wg socket in the host after moving the interface into a child vnet. Working on a test for that too.
Excellent, thanks! It's currently a little broken trying to connect to external hosts, but it smells like a timer problem. tcpdump shows packets going out and packets coming back in at a rapid rate and you can observe debug messages to the console even about handshake and keepalive packets flowing.
| tests/sys/netinet/if_wg_test.sh | ||
|---|---|---|
| 88 | After playing around with it a little bit, I'd actually recommend -o instead of -c 3 and also -i 0.25 if we can do that reliably. | |
| tests/sys/netinet/if_wg_test.sh | ||
|---|---|---|
| 88 | Actually, we can probably drop the timeout entirely and just use -o. See the commit upstream about adding a stage queue -- we were consistently dropping the first icmp packet, but we shouldn't have been; that likely applies here, too. | |
No, I was following the example of other cloned interface tests (see tests/sys/net) and tried to be consistent with the majority. I will change it.
Got it. I don't have a strong opinion. It's more curious. There are two conventions of naming in tests/sys/net so I think both of them are fine. Eventually, we should unify the naming, but that can wait after our tests get more mature.
Oh no problem. Was it worth committing at all in light of the port of netns.sh? I didn't spend much more time on this once I saw you were working on that.
I think it's useful to have some basic functional tests hooked up for Jenkins to bristle at -- netns.sh covers all of this, but it's a lot more like stress2 in nature