Open abradshaw opened 6 years ago
Maybe also server loadbalancing aka "balanced tandem" via DNS TXT/SRV records
Maybe also repository discovery (site specific) via DNS TXT/SRV.
eg.: myrepository._mysite._yum.example.com -> "https://biycle.mysite.example.com/repos/myrepository"
Implement site discovery as well. This automatically points the client to the next repo available.
Example:
[luc@f28 ~]$ dig _pkgrepo._tcp.example.com. SRV _pkgrepo._tcp.example.com. 86400 IN CNAME _ldap._tcp.NUE._locations.example.com. _pkgrepo._tcp.BER._locations.example.com. 86400 IN SRV 0 100 80 example.com.
Where 0 is the prio, 100 the weight and 80 the port.
Prototype needed....
Let us do the resolving part on the bicycle server and set the URL in the client repo file. So we have no dependencies on the clients, like scripts or yum plugins. Simple wins.
Downside, how to discover the site a client belongs to?
Could use DNS lookup function from ansible or run "dig" via shell.
Locate preferred bicycle server via simply querying for a DNS SRV record