On 2017-01-11 15:01, Pete Wright wrote: > > > On 1/11/17 11:55 AM, Allan Jude wrote: >> On 2017-01-11 14:33, Jason wrote: >>> Hi, >>> I installed vnstat, which is a bandwidth monitoring app. I noticed >>> that >>> for some reason its TX bytes were 0, but the RX bytes were being >>> tracked as >>> expected. I don't know how vnstat is collecting its metrics every 5 >>> minutes, but I'm guessing it's similar to how 'netstat -I xn0 -bn' would >>> collect its details. After running 'netstat -I xn0 -bn', I notice that >>> "Obytes" for "<Link#2>" is 0, which may be the problem. However, the >>> "Obytes" for the actual IP address seems to be working: >>> >>> # netstat -I xn0 -bn >>> Name Mtu Network Address Ipkts Ierrs Idrop >>> Ibytes Opkts Oerrs Obytes Coll >>> xn0 1500 <Link#2> <REDACTED> 22659117 0 0 >>> 2579194762 >>> 21221967 0 0 0 >>> xn0 - 10.0.10.0/24 10.0.10.40 22567359 - - >>> 2258113267 >>> 21130240 - 3518207522 - >>> >>> On a different machine that command works fine. It is running BSD 11 as >>> well, but it's not using the xn driver -- it uses the fxp driver. It's >>> output is: >>> >>> # netstat -I xn0 -bn >>> Name Mtu Network Address Ipkts Ierrs Idrop >>> Ibytes Opkts Oerrs Obytes Coll >>> em0 1500 <Link#1> <REDACTED> 1507204306 0 0 >>> 1567750969110 100389817 0 80619546746 0 >>> em0 - 192.168.10.0/ 192.168.10.10 1506281038 - - >>> 1548639235291 100283723 - 79211140098 - >>> >>> I'm running FreeBSD 11 on both instances, except the one that's having >>> difficulty is running on AWS, so is on Xen. >>> >>> Any insight into how to fix this so "Obytes" populates for "Link#2"? >>> >>> Thanks! >>> _______________________________________________ >>> [email protected] mailing list >>> https://lists.freebsd.org/mailman/listinfo/freebsd-cloud >>> To unsubscribe, send any mail to "[email protected]" >>> >> >> There is a fix for this here: >> >> https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=213439 > > Oh great - do you think this will also address this PR: > https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=213814 > > If I have time this week I'll try to verify this as well and will close > 213814 if it works. > > Cheers, > -pete >
No, it does not fix the ixv(4) problem. It was a bug specifically in xn(4), which is the paravirtual nic. ixv is an SR-IOV passthru device. So it is a different problem. However, head just got an all new framework for the intel NIC drivers, I am wondering if THAT fixes this issue, likely by coincidence. -- Allan Jude
signature.asc
Description: OpenPGP digital signature
