send: No buffer space available for ppp interface

[Luis Mondesi]

Binary package hint: traceroute

when trying to traceroute using a ppp interface, the tx_queue_len buffer fills up and causes this weird alert.

this is a known issue that has been fixed on Fedora: http://webui.sourcelabs.com/fedora/issues/245438

this works:

sudo traceroute -N 7 foo.bar

but this does not:

sudo traceroute foo.bar
traceroute to foo.bar ...
send: No buffer space available

Please use the fedora patch and fix this puppy

dpkg --list traceroute
ii traceroute 2.0.9-3 ...

Hardy 8.04.1

[TimWintle (timwintle)]

Confirmed (I'm having the same issue on Hardy 64-bit over a 3G ppp connection)

Affects both traceroute and tcptraceroute

sample command line output:

(as standard user)
$ tcptraceroute launchpad.net
The specified type of tracerouting is allowed for superuser only

(with superuser privileges)
$sudo tcptraceroute launchpad.net
traceroute to launchpad.net (91.189.90.211), 30 hops max, 40 byte packets
send: No buffer space available

The link in the original post includes a bug report from a kubuntu user.

Setting status to Confirmed.

[Graham Inggs (ginggs)]

Fix released in 2.0.10-1 in intrepid.

Upstream changelog entry:
2008-04-17 Dmitry Butskoy <email address hidden> - 2.0.10

 * Interpret ENOBUFS errors for send(2) as "can retry later".

    Slow devices (like ppp) with small tx_queue_len can reject
    the sending of too many packets simultaneously. To handle this,
    do_send() now returns a negate value in a case of ENOBUFS
    and similar (instead of program exit). The send_probe method
    clears the probe and returns immediately in such cases.
    Then, if there is an amount of time to wait for something,
    the failed probe will be attempted again after that time expired.
    If nothing to wait more, the program is exited.