So today I had to restart my portmapper (bizarre fact of life) and after
restarting it and all the services running, I noticed that locking was
broken. The reason was that clients no longer could contact the lockd,
since it wasn't registered after we reran the portmapper.
Now that would be fixable if we could tell lockd to restart, or kill it
and restart it manually, but I don't know how to do it. I had to reboot
the server.
Is there a way to kill lockd, or to tell it to restart and register
again in the portmapper? I do know that something (or maybe it's
suicide) tries to wait for lockd to die upon shutting down on our
diskless network, without success. Any hints?
Take care,
--
Christian Reis, Senior Engineer, Async Open Source, Brazil.
http://async.com.br/~kiko/ | [+55 16] 261 2331 | NMFL
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs
>>>>> " " == Christian Reis <[email protected]> writes:
> Is there a way to kill lockd, or to tell it to restart and
> register again in the portmapper? I do know that something (or
> maybe it's suicide) tries to wait for lockd to die upon
> shutting down on our diskless network, without success. Any
> hints?
Why can't you use pmap_dump + pmap_set? The above type of recovery
situation is exactly what they have been developed for.
Alternatively, you would have to umount all the partitions that are
running without 'nolock' set.
Cheers,
Trond
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs
On Thu, Aug 28, 2003 at 02:58:23PM -0400, Trond Myklebust wrote:
> >>>>> " " == Christian Reis <[email protected]> writes:
>
> > Is there a way to kill lockd, or to tell it to restart and
> > register again in the portmapper? I do know that something (or
> > maybe it's suicide) tries to wait for lockd to die upon
> > shutting down on our diskless network, without success. Any
> > hints?
>
> Why can't you use pmap_dump + pmap_set? The above type of recovery
> situation is exactly what they have been developed for.
>
> Alternatively, you would have to umount all the partitions that are
> running without 'nolock' set.
Hmmm. This isn't on a client, though, it's on a server - it doesn't
mount any foreign exports. Or did you literally mean partitions?
Now, pmap_dump/_set would have helped *if* I had known about them
previously to killing the portmapper <wink>
Take care,
--
Christian Reis, Senior Engineer, Async Open Source, Brazil.
http://async.com.br/~kiko/ | [+55 16] 261 2331 | NMFL
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs
>>>>> " " == Christian Reis <[email protected]> writes:
>> Alternatively, you would have to umount all the partitions that
>> are running without 'nolock' set.
> Hmmm. This isn't on a client, though, it's on a server - it
> doesn't mount any foreign exports. Or did you literally mean
> partitions?
If it's on the server, then you just need to kill the nfsds.
> Now, pmap_dump/_set would have helped *if* I had known about
> them previously to killing the portmapper <wink>
The pmap_dump format for files is the same as that returned by
rpcinfo. i.e. it is human-readable and can therefore be reconstructed
if you know which daemons are running on which ports.
'netstat -ap' is your friend (as is 'lsof').
Cheers,
Trond
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs