On Tue, 2020-06-23 at 12:32 -0500, Patrick Goetz wrote: > Hi - > > Sorry for the delayed response. Work has really picked up the last > couple of weeks. > > On 6/17/20 8:28 PM, Ian Kent wrote: > > > Maybe this is just an issue with mount.nfs? > > > > It's hard to say, possibly, that would need a debug log to check it > > out. > > Although the problem is completely resolved by specifying > > -tcp,vers=4.2 > > in /etc/auto.home, I'm happy to turn on debugging on one of the > workstations and send the results if this would help. The only thing to consider is what's in the distro autofs revision and what patches may have been added. The key thing is to not attempt a mount if, for whatever reason, the server doesn't respond and as long as autofs works this out rather than letting mount.nfs(8) do it, it should be fine. The server probing done by autofs deliberately uses quite short timeouts so things don't get stuck. And your installed autofs appears to have the the code that ensures autofs itself does this check so it should be what's working out whether to attempt a mount or not fairly quickly. So I'm not sure we need to do more unless there are further problems. > > > > If the problem is bad enough it might be worth adding some entries > > above the wildcard entry that bind mount to one or more directories > > to keep those applications happy. > > > > Or create dummy home directories on the NFS server, or change > /etc/passwd to the more correct: > > /home/syslog --> / > > however, given that, for example, > > $ ls /home/syslog > > returns instantly with "no such file or directory" when the mount > options -tcp,vers=4.2 are specified in /etc/auto.home, my problem is > functionally resolved; I'm just in the bug reporting stage. So, as long as your happy to specify those options everything should fine. If you wanted autofs to automaticly work out (without those options) what to do it gets quite a bit harder, particularly when using a wildcard map entry. That's because the map entry key will always match the wildcard entry so there's no way to know if the entry is actually valid so that mount attempts can be avoided. That was the point of the suggestion of adding particular problem entries above the wildcard entry, if needed. Ian
