Ilya Dryomov <idryomov@xxxxxxxxx> writes: > On Mon, Dec 14, 2020 at 4:55 PM Luis Henriques <lhenriques@xxxxxxx> wrote: >> >> Ilya Dryomov <idryomov@xxxxxxxxx> writes: >> >> > Hello, >> > >> > I've pushed wip-msgr2 and opened a dummy PR in ceph-client: >> > >> > https://github.com/ceph/ceph-client/pull/22 >> > >> > This set has been through a over a dozen krbd test suite runs with no >> > issues other than those with the test suite itself. The diffstat is >> > rather big, so I didn't want to spam the list. If someone wants it >> > posted, let me know. Any comments are welcome! >> >> That's *awesome*! Thanks for sharing, Ilya. Obviously this will need a >> lot of time to digest but a quick attempt to do a mount using a v2 monitor >> is just showing me a bunch of: >> >> libceph: mon0 (1)192.168.155.1:40898 socket closed (con state V1_BANNER) >> >> Note that this was just me giving it a try with a dummy vstart cluster >> (octopus IIRC), so nothing that could be considered testing. I'll try to >> find out what I'm doing wrong in the next couple of days or, worst case, >> after EOY vacations. > > Hi Luis, > > This is because the kernel continues to default to msgr1. The socket > gets closed by the mon right after it sees msgr1 banner and you should > see "peer ... is using msgr V1 protocol" error in the log. > > For msgr2, you need to select a connection mode using the new ms_mode > option: > > ms_mode=legacy - msgr1 (default) > ms_mode=crc - crc mode, if denied fail > ms_mode=secure - secure mode, if denied fail > ms_mode=prefer-crc - crc mode, if denied agree to secure mode > ms_mode=prefer-secure - secure mode, if denied agree to crc mode Ah, right. I should have took a quick look at the patches first to check for any new parameters. Thanks for pointing me at that, I'll retry my test using ms_mode. Cheers, -- Luis