I am trying to figure out why our publicly reachable schedd nodes are consistently picking the wrong network interface.
Â Â Â Â The only address changed by ENABLE_ADDRESS_REWRITING is the address that's /advertised/ to the collector.Â Unless you turn off BIND_ALL_INTERFACES, the daemons actually listen on all addresses.Â What they report in their log is the address they would advertise without rewriting.Â Since the only daemon which is contacted without looking it up in the collector is the collector (or the shared port daemon, which will impersonates the collector), changing a daemon's address in the collector is (almost always) all that's necessary; making this change /only/ in the collector allows a daemon on a multi-homed machine to advertise an appropriate address in whichever collector(s) it contacts.
Â Â Â Â If you change NETWORK_INTERFACE back to *, does 'condor_status -schedd -af MyAddress' return public or private IPs?Â If private, could you add D_NETWORK:2 to your SCHEDD_DEBUG, start up a schedd and send me its log?Â Thanks.
HTCondor-users mailing list
To unsubscribe, send a message to htcondor-users-request@xxxxxxx
You can also unsubscribe by visiting
The archives can be found at: