Il giorno ven, 09/03/2012 alle 14.40 -0600, Andrew Deason ha scritto: > Everything you've said so far makes it sound like _something_ is getting > dropped along the way, due to the rather broad nature of the > problems/timeouts. And I take it the network layout you have for this is > not exactly the simplest thing, is that correct? There's some firewalls > and VPNs and maybe NATs getting traversed; is that right? My guess would > be fragmented packets are getting lost, or maybe just packets over a > certain size.
You're absolutely right. The net setup is that: afsmn1, afsmn3, afsmn4, ..: - bridge (KVM virtualization) over - vlan (virtualizer host, vlan-N) over - bond (virtualizer host, eth0 + eth1 + ..) over - ethernet (virtualizer host) - local net - OpenVPN (on a local gateway machine) - afsrm1 (OpenAFS + OpenVPN) Maybe too mani levels of packet encapsulation? > In addition to all that, you can run a more 'real' test of connectivity > between the servers with rxperf. If you don't have rxperf, you can build We tested a bit with a very simple method: netcat (we'll try rxperf in the future, when things are a bit more "calm"): afsrm1: netcat -t -l -p 8000 -v -v -n afsmn1: netcat -u afsrm1 8000 Just chatting a bit between the 2 hosts we noticed that many messages were lost.. After that we tried to introduce in our openvpn.conf files these lines, in order to reduce effective packet size: mssfix 1200 fragment 1200 That did the trick! :) We are testing that in 5 days now and everything seems ok! Fast communications, no more timeouts or failures during releases, correct ptdb/vldb propagation.. GOOD! :) We hope this can help also other OpenAFS+OpenVPN users! Now we have one last question about our setup: in afsrm1 (the remote server) we used /var/lib/openafs/local/NetInfo (debian locations): --- 192.168.11.12 f 10.250.10.11 --- Is that correct? Note: 192.168.11.0/24 is the remote subnet and 10.250.10.11 is the tun0 IP (openvpn). We see afsrm1 as a multihome (in VLLog/Ptlog, with udebug, etc..). Is that a good or bad thing? Thank you again for all!! Stefano Fabio _______________________________________________ OpenAFS-info mailing list [email protected] https://lists.openafs.org/mailman/listinfo/openafs-info
