Re: -current network buffer exhaustion on RPi2 armv7

On Mon, Mar 09, 2026 at 08:16:58AM -0700, bob prohaska wrote:
> On Sun, Feb 22, 2026 at 09:41:41AM -0800, Adrian Chadd wrote:
> > I'm telling ya, it's dhclient, trying to send frames when the
> > interface is up but encryption isn't finished negotiating.
> 
> It appears that leaving a ping session running with an interval
> of ten seconds somehow suppresses the network connection drop
> and wpa_cli chatter.
> 

Well, maybe not. Merging the output of ping, the console 
and a poudriere bulk build there just appeared:

...
64 bytes from 50.1.20.27: icmp_seq=5810 ttl=63 time=4.010 ms
64 bytes from 50.1.20.27: icmp_seq=5811 ttl=63 time=4.558 ms
64 bytes from 50.1.20.27: icmp_seq=5812 ttl=63 time=7.111 ms
swap_pager: indefinite wait buffer: bufobj: 0, blkno: 74939, size: 20480
swap_pager: indefinite wait buffer: bufobj: 0, blkno: 127635, size: 4096
ping: sendto: Network is down
Mar  9 11:35:24 generic wpa_supplicant[3380]: ioctl[SIOCS80211, op=20, val=0, arg_len=7]: Can't assign requested address
ping: sendto: No route to host
ping: sendto: No route to host
64 bytes from 50.1.20.27: icmp_seq=5816 ttl=63 time=4.641 ms
64 bytes from 50.1.20.27: icmp_seq=5817 ttl=63 time=7.443 ms
64 bytes from 50.1.20.27: icmp_seq=5818 ttl=63 time=5.468 ms
64 bytes from 50.1.20.27: icmp_seq=5819 ttl=63 time=6.635 ms
...

Evidently the pings didn't prevent the disconnect. Perhaps they
contributed to re-connection in about forty seconds. Usually the
connection stays down once dropped. 

The "indefinite wait..." messages are normal when a Pi is swapping
actively. The wpa messages can be seen at any time, loaded system
or idle. Seeing them consecutively seems a bit odd...

Thanks for reading,

bob prohaska
 

 

 
>