Closed 1977er closed 2 years ago
Trying to increase net.ipv6.route.max_size
according to https://www.mail-archive.com/bird-users@network.cz/msg05945.html.
Setting /etc/sysctl.d/11-ipv6-routing.conf
manually.
Doesn't help against Kernel dropped some netlink messages, will resync on next scan
messages.
Monitoring route to bgp endpoints via Zabbix.
Adjusting net.ipv6.route.max_size
might have helped against No route to host
. No hold-timer issues since then. Will keep an eye on it anyway.
I guess the sysctl setting did the trick. No more No route to host
for three days now. I declare the patient cured.
Moving the zabbix debug items into the template, but disabling them.
Since (around) 01/2022 bird is flapping ipv6 bgp sessions (both cix1 and cix2).
Logs show alot of
clevertor bird[750]: Kernel dropped some netlink messages, will resync on next scan.
andclevertor bird[750]: Netlink: No route to host
before finallyclevertor bird[750]: cix1_ip6: Error: Hold timer expired
In the aftermath routes from the ipv6 table get deleted (assumption) and ipv6 connectivity on some host (sn02,...) go south (fact).