On Wed, 2015-08-19 at 21:24 +0200, Olaf Hering wrote:
After upgrading from openSUSE 11.4 with NM 0.9.something to openSUSE Tumbleweed with NM 1.0.2 I was under the impression that the interface/API for the scripts in /etc/NetworkManager/dispatcher.d is much more reliable, and much more informative. But just now I got a different interface for the up and the down event of a gsm connection. Thankfully the DEVICE_IFACE is the same for the up and down event, so I will adjust my scripts to use that.
Probably that is expected behavior, because of http://cgit.freedesktop.org/NetworkManager/NetworkManager/tree/callouts/nm-dispatcher-utils.c?id=1c03d92f4db80dc46fc0c4380541b541bfa3576d#n451
On another host I get different environment for initial vpn-up and a vpn restart. The latter lacks the route info because that is not provided during restart after the vpn gateway timed out. Right now I dont have the exact env data at hand, but I have added logging now to compare both variants. Perhaps there are already ways to catch both. I wonder if there is any promise for the called scripts, or if its just more or less random data that is provided as cmdline arguments and environment.
I think the dispatcher scripts aim to be as reliable as possible. And to behave as documented in `man NetworkManager`. Maybe it'd be best, if you point out where you find: - `man NetworkManager` being unclear, inaccurate, or wrong. - which data you actually find wrong/unexpected, or contradicting the the manual. Thomas
Attachment:
signature.asc
Description: This is a digitally signed message part