Carlos E. R. wrote:

El 2013-10-10 a las 18:27 +0530, AP escribió:

...

not in that, it won't break the system except to restart! However, it might be due to the olden days called backward compatibility like they call it.

If you restart a machine exporting directories via NFS, clients may crash. It has happened to me.

Only if you have the systems set up wrong. One of the big advantages of NFS is that the server can go away and then come back and running processes on clients will simply carry on, without having to have any awareness of it at all. (i.e. the code is not aware of the NFS mount or any errors etc). I do it/it happens to me regularly. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos E. R. wrote:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256

El 2013-10-11 a las 09:45 +0100, Dave Howorth escribió:

...

...

If you restart a machine exporting directories via NFS, clients may crash. It has happened to me.

Only if you have the systems set up wrong.

Please explain that, I'm interested.

I can't really tell what setting on your machine is wrong. But see below...

...

One of the big advantages of NFS is that the server can go away and then come back and running processes on clients will simply carry on, without having to have any awareness of it at all. (i.e. the code is not aware of the NFS mount or any errors etc). I do it/it happens to me regularly.

If the server remains off (because it is intended to be off, day end), the client had to be rebooted to continue.

No, the client will sit there until the server comes back next day. That is how it is designed. And certainly the client hasn't crashed at that point. So have you had a crash or not? I'm not sure what behaviour you want your particular client to have. But the most obvious change would be to configure the mount with the soft option. Then your client would have to be aware of possible errors (as it should be in any case of course). When the server goes down, the client will receive an error and it's up to the client to deal with that. NFS is very like TCP. It assumes that breaks in the network, or servers going away, are temporary phenomena and just waits until conditions return to normal when it carries on. And NFS, like TCP, can be told to return errors instead, if that's what you want. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos E. R. wrote:

On 2013-10-11 12:21, Dave Howorth wrote:

...

No, the client will sit there until the server comes back next day. That is how it is designed. And certainly the client hasn't crashed at that point. So have you had a crash or not?

The nfs was mounted as /data/external/. The entire /data directory was unaccessible, any process trying hung. Mind, not accessing /data/external/, but /data/. Several processes got stuck, things failed. The machine had to be rebooted to be usable again. mount/umount hung, impossible to remove the nfs mount.

Something weird was happening there. If it's a reproducible, ongoing problem, we can try to solve it, but otherwise trying to draw any conclusions from the circumstances is very error-prone.

Both server and client were setup by yast, so any misconfiguration was due to yast.

I did have a server go down yesterday and after it came back there were odd symptoms when starting new processes on the clients until I went into YaST and restarted the NFS server component. But existing processes weren't affected at all. I don't understand what happened there, and I probably never will.

...

I'm not sure what behaviour you want your particular client to have. But the most obvious change would be to configure the mount with the soft option. Then your client would have to be aware of possible errors (as it should be in any case of course). When the server goes down, the client will receive an error and it's up to the client to deal with that.

Both are 12.3 machines. So you say Linux is buggy... :-p

Sorry, I don't understand that logic, or how your statement relates to what I said. Cheers, Dave -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Ruediger Meier wrote:

I guess that most people who wants to reboot machines have also the root password. So they would not need this permission for their user accounts. Giving all non-administrators the reboot permission per default is wrong. If root wants to give them reboot permission then he could do it via sudoers. This console/remote user magic is IMO useless.

I have often used ssh to connect to a computer and then, as root, rebooted or halted it. You don't have to be sitting in a box to reboot or halt. You don't even have to be root, if you use the desktop or menu to halt or restart. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

El 11/10/13 09:59, Ruediger Meier escribió:

That's exactly what I complain about.

Looks like you have different requirements, particularly it seems you are expecting something that behaves like an old unix or a distribution from the 90's.. those are dead and buried in case you haven't noticed. -- "If debugging is the process of removing bugs, then programming must be the process of putting them in." - Edsger Dijkstra -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Le 11/10/2013 17:45, Bernhard Voelker a écrit :

I also think the current shutdown/reboot permissions are too lax.

may be Yast / security / hardening? but in the pastr some applications could have problems with hardening. jdd -- http://www.dodin.org -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org