I have been struggling to find the reason that my PPP link
locks up about every 6-12 hours with "no buffer space" messages.

I am beginning to suspect that it is related to the number of
fragmented packets that go over the link.

Can someone on this list confirm or deny this behaviour?

Could someone offer a suggested rsize/wsize etc. that 
might help the following stats. Reducing fragments?


Info from netstat -s

ip:
        3540021 total packets received
        0 bad header checksums
        0 with size smaller than minimum
        0 with data size < data length
        0 with header length < data size
        0 with data length < header length
        0 with bad options
        0 with incorrect version number
        856387 fragments received
        0 fragments dropped (dup or out of space)
        180 fragments dropped after timeout
        50325 packets reassembled ok
        2460011 packets for this host
        273211 packets for unknown/unsupported protocol
        712 packets forwarded
        0 packets not forwardable
        0 redirects sent
        2692589 packets sent from this host
        0 packets sent with fabricated ip header
        0 output packets dropped due to no bufs, etc.
        225922 output packets discarded due to no route
        910 output datagrams fragmented
        3964 fragments created
        0 datagrams that can't be fragmented