Hello Andrey and FreeBSD Net, Today I stumbled upon what may be zero, one, or maybe two bugs. If it's one bug, then I think there might be a case the fine work in r339552 [0] to keep the RUNNING flag consistent with whether or not the gre source address exists does not accommodate. I'm still working on a minimal test case, so bear with me.
If I create gre0 on the host, then start a VNET jail which gets gre0 and a physical interface (msk1, in my case) and inside which both are configured, all is well inside the jail. Upon stopping the jail, the "tunnel" setting is not removed from the gre (I'm not sure whether to consider this a bug, too, or not), so when the dust settles back on the host, msk1 has no address, and the gre is not RUNNING, which is fine. However, if I start the jail again, then the gre is not RUNNING inside the jail even though msk1 again has the appropriate source address. I worked around this by creating gre0 inside the jail where it is destroyed upon jail stop. It would be nice if it either worked the other way, or failed with less astonishment. Am I missing something that should be obvious? If this does seem like a bug, shall I submit something now or wait until I have a cleaner demo? [0] https://svnweb.freebsd.org/changeset/base/339552 (thank you!) Regards, -- John W. O'Brien OpenPGP keys: 0x33C4D64B895DBF3B
signature.asc
Description: OpenPGP digital signature
