On Thu, Nov 19, 2009 at 01:46:46AM +0000, Stuart Henderson wrote: > a couple of things I think it really needs: > > - allow the socket to be specified (bgpctl -r), for people who want > to use the r/o socket or people running multiple daemons
Done. What I didn't do was allow you to specify a different socket for different peers, it is just a single socket for the entire check. It would be possible any not too difficult to add, but for now, multiple daemons can be checked with multiple checks. Is the ability to specify multiple sockets important? You could: check_openbgpd -s daemon1.rsock -w x:y -c a:z -n P1 -s daemon2.rsock -n P2 But, describing what happens with that check is tough. What about -u? Does it work per-socket or globally? What if P2 appears on daemon1.sock? Is that OK? > - handle the output format with max-prefix in use e.g. > > testpeer 41103 403389 34611 0 01:17:35 > 299345/500000 Fixed. > also I have a short wish-list if you're interested and have time > to add more features: > > - option to change the alarm given when a peer exists in the > bgpctl output but not on the command line (none/warn/critical) Added (sort of) You can now specify -u at any point to mean unknown peers. You can specify it first to disable all checks of unknown peers (which is like none) or you could specify -c 'UNKNOWN' -u which will error with WARNING that the peers state is not an acceptable value. See the examples in help. > - alarm if > X% of max-prefix paths are received. e.g. warn if you > get 75/100, critical if you get 90/100. very useful for IX-facing > routers Added, for any peers with max-prefix it should work, for peers without it errors about no max-prefix. > and it would be really nice if this could work automatically > for all peers, rather than having to list them one-by-one. You can now "check_openbgpd -w :75% -c :90% -u" which should do what you describe. It will tho also be critical if state is non-numeric or no max-prefix. I don't know how you could specify a % and not have a neighbor without max-prefix be an error. > - maybe a check for nexthops too...I imagine it would be saner > done as a separate script though. Not added, lots more work, but I will think about how to do it. l8rZ, -- andrew - ICQ# 253198 - Jabber: and...@rraz.net BOFH excuse of the day: Post-it Note Sludge leaked into the monitor.
check_openbgpd-1.5-port.tar.gz
Description: application/tar-gz