Comment 6 for bug 1936907

Revision history for this message
Didier Roche-Tolomelli (didrocks) wrote :

Thanks Seth for the review and the overall positive comments! :)

Some answers:
1. the potential race is fixed after our discussion and pending some reviews

2. the pam modulefixes are done and merged already (even if upstream don’t deallocate, let’s do it on our side)

3. on the conditions that can be added to adsys-boot.service to make it less likely to spam the journal every five seconds for ten hours when on an airplane?
-> We can’t rely on network being up (maybe we never had the network, or the interface is on but not connected yet, or the interface is on, has no Internet, but local network is enough to reach AD).
Depending on all those conditions, we can’t link it to the network, it may be too early or too late. Also, we support offline mode once we have a valid cache.

Considering that this case only happen the first time you boot your machine (no local cache for offline usage) and don’t have access to AD, this doesn’t seem a big issue and rather something you want to be alerted on, what do you think?

4. on the doc and examples containing a socket in /tmp
-> This is more a debug example to run adsysd as non root. The issue with putting real values is then, if you do that on a system where adsysd is running, you end up erroring out on the systemd existing socket and then, it’s a nightmare to recover on the systemd side (you need to reset the state of the .socket unit). This is why the example carefully avoid using the real system socket (in addition to require root to read it).