On 2018/12/18 11:34, Arnaud BRAND wrote: > Hi, > > I'm running 6.4 stable, with latest syspatches. > > I saw ospf6d reporting this in the logs > Dec 18 08:18:10 obsd64-ic1 ospf6d[68658]: send_packet: error sending packet > on interface vmx1: No buffer space available > > Searching the web, I gathered that netstat -m might shed some light, so I > proceeded : > obsd64-ic1# netstat -m > 610 mbufs in use: > 543 mbufs allocated to data > 8 mbufs allocated to packet headers > 59 mbufs allocated to socket names and addresses > 13/200 mbuf 2048 byte clusters in use (current/peak) > 0/30 mbuf 2112 byte clusters in use (current/peak) > 1/56 mbuf 4096 byte clusters in use (current/peak) > 0/48 mbuf 8192 byte clusters in use (current/peak) > 475/2170 mbuf 9216 byte clusters in use (current/peak) > 0/0 mbuf 12288 byte clusters in use (current/peak) > 0/0 mbuf 16384 byte clusters in use (current/peak) > 0/0 mbuf 65536 byte clusters in use (current/peak) > 10196/23304/524288 Kbytes allocated to network (current/peak/max) > 0 requests for memory denied > 0 requests for memory delayed > 0 calls to protocol drain routines > > So if there were no requests denied or delayed and the peak was only 24MB > out of 512MB max, what could cause ospf6d to complain ? > Should I be worried about this message ? > > Looking at the sendto man page I get that it can return ENOBUFS in two cases > : > Case 1 - The system was unable to allocate an internal buffer > -> this seems to not be the case as shown above > > This leaves only case 2 : The output queue for a network interface was full. > > Looking at netstat -id is see drops on vmx1 and vmx3. > Both of these cards are VMXNET3 cards connected to the different > VLANs/Portgroups on the same vswitch which has two 10G uplinks to the > switches. > > sysctl | grep drops shows > net.inet.ip.ifq.drops=0 > net.inet6.ip6.ifq.drops=0 > net.pipex.inq.drops=0 > net.pipex.outq.drops=0 > > I'm out of ideas for places where to look next. > Please, could network guru provide some insight/help ? > Or just tell me that it's not worth bothering and I should stop here ? > > Thanks for your help and have a nice day ! > Arnaud
It maybe worth trying e1000/em(4). I had quite frequent panics with vmx(4) (https://marc.info/?l=openbsd-bugs&w=2&r=1&s=vmxnet3_getbuf&q=b), the same VM has been totally stable since switching to em(4).
