Closed from88 closed 5 years ago
assign me, I'll debug more what's happening here.
[root@exit1-centos-76 frr]# ps aufx | grep frr
root 6328 0.0 0.0 112708 980 pts/0 S+ 19:40 0:00 \_ grep --color=auto frr
frr 6174 0.0 0.3 355564 3368 ? S<sl 19:34 0:00 /usr/lib/frr/zebra -d -A 127.0.0.1
frr 6183 0.0 0.4 210580 4756 ? S<sl 19:34 0:00 /usr/lib/frr/bgpd -d -A 127.0.0.1
frr 6200 0.0 0.3 60456 3200 ? S<s 19:34 0:00 /usr/lib/frr/ospfd -d -A 127.0.0.1 -n 1
frr 6209 0.0 0.2 60324 2696 ? S<s 19:34 0:00 /usr/lib/frr/ospfd -d -A 127.0.0.1 -n 2
frr 6219 0.0 0.1 58820 1796 ? S<s 19:34 0:00 /usr/lib/frr/staticd -d -A 127.0.0.1
root 6233 0.0 0.1 58392 1432 ? S<s 19:34 0:00 /usr/lib/frr/watchfrr -d -r /usr/lib/frr/frr restart %s -s /usr/lib/frr/frr start %s -k /usr/lib/frr/frr stop %s zebra bgpd ospfd-1 ospfd-2 staticd
vtysh is trying to connect to ospfd.vty
.
[root@exit1-centos-76 frr]# strace -econnect -ff -s2048 vtysh -c 'conf t' -c 'router ospf 2' 2>&1 | grep ospfd
connect(4, {sa_family=AF_LOCAL, sun_path="/var/run/frr/ospfd.vty"}, 24) = -1 ECONNREFUSED (Connection refused)
connect(4, {sa_family=AF_LOCAL, sun_path="/var/run/frr/ospfd-1.vty"}, 26) = 0
connect(5, {sa_family=AF_LOCAL, sun_path="/var/run/frr/ospfd-2.vty"}, 26) = 0
Warning: connecting to ospfd...connect(9, {sa_family=AF_LOCAL, sun_path="/var/run/frr/ospfd.vty"}, 24) = -1 ECONNREFUSED (Connection refused)
Basically starting up the main ospfd process without instance specified additionally, would solve the problem (trivial changes to frr.in
). But there will be a way to configure router ospf
and router ospf 1
both, which is confusing.
[root@leaf1-centos-76 frr]# git bisect bad
67736451c55347849909cb3a05c706d5f9e2c29a is the first bad commit
commit 67736451c55347849909cb3a05c706d5f9e2c29a
Author: Mladen Sablic <mladen.sablic@gmail.com>
Date: Mon May 21 20:00:51 2018 +0200
vtysh: reconnect to daemons when connection lost
Functionality to let vtysh attempt to reconnect to daemons when
connection is lost (e.g. crash or restart).
Signed-off-by: Mladen Sablic <mladen.sablic@gmail.com>
:040000 040000 cc3deb582d09a45b7924611aecd914738f96b8f6 cd70810673bc305406e30f655c907e470078fbec M vtysh
@ton31337 do you have b2443937b06d79f1caa46cdb7625e3d6831b2166 in the non-working case?
yeah, this fixes the problem entirely. we should drop https://github.com/FRRouting/frr/commit/b2443937b06d79f1caa46cdb7625e3d6831b2166 to 6 branch :)
Also, here's a branch converting RPMs over to the new init scripts: https://github.com/opensourcerouting/frr/tree/6.0/redhat-new-init
NB: UNTESTED.
tested this - works with the applied change in stable/6.0 (https://github.com/FRRouting/frr/commit/b2443937b06d79f1caa46cdb7625e3d6831b2166)
after installing RPM package for CENTOS, multiple OSPF instances doesn't work:
With the help of equinox (On slack) we tried to workaround this:
We tried to make the service run with: ExecStart=/usr/lib/frr/frrinit.sh start also add options: watchfrr_options="-r '/usr/lib/frr/watchfrr.sh restart %s' -s '/usr/lib/frr/watchfrr.sh start %s' -k '/usr/lib/frr/watchfrr.sh stop %s'"
but after that i got some other errors:
i'm using CentOS Linux release 7.6.1810 (Core).
Thanks