2002-10-08 14:29:12

by tswaine

[permalink] [raw]
Subject: NFS server replacement troubles

I currently have a SCO NFS server, however I'm replacing it with Redhat
Linux.
I have taking down the box replaced it and NFS fails. The hostname, ip, etc.
is
the same on the new machine.
New box info....
Stock Redhat Linux 6.2
kernel 2.2.14-6.0.1 on i686
partition being exported is Reiser filesystem

Before new box was put in to production its hostname was different.
Exporting was fine, NFS worked properly, however as soon as box
is brought up under new name NFS fails, errors(see below ).

Snapshot of errors.......

Starting portmapper: [ OK ]
master:/var/lib/nfs 23# /etc/rc.d/init.d/nfs start
Starting NFS quotas: Cannot register service: RPC: Unable to receive; errno
= Connection refused
[FAILED]
Starting NFS mountd: Cannot register service: RPC: Unable to receive; errno
= Connection refused
mountd: unable to register (mountd, 1, udp).
[FAILED]
Starting NFS daemon: nfssvc: Connection refused
[FAILED]
Starting NFS services: [ OK ]


I have restarted portmapper and nfs with no success, unmount nfs on all
connected machines, still no workee....
Any help would be appreciated...
Thanks.


Troy L. Swaine
System Administrator / Engineer



-------------------------------------------------------
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


2002-10-08 16:27:49

by Benjamin LaHaise

[permalink] [raw]
Subject: Re: NFS server replacement troubles

On Tue, Oct 08, 2002 at 10:28:52AM -0400, [email protected] wrote:
> Stock Redhat Linux 6.2
> kernel 2.2.14-6.0.1 on i686
> partition being exported is Reiser filesystem

That kernel is utterly ancient. At the very least you should upgrade to the
newer errata that were released for 6.2 to fix security issues in the kernel
and userland, but preferably you should just replace the box with one running
7.3, as 6.2 is pretty much end of life. Numerous bugs in NFS, reiserfs and
the core kernel have been fixed since 2.2.14 was released almost 2 years ago.

-ben


-------------------------------------------------------
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

2002-10-08 17:10:20

by tswaine

[permalink] [raw]
Subject: RE: NFS server replacement troubles

Thanks for your reply, but I have to use 6.2 with that
kernel for my specific application. It short it seems the
system works till I replace the hostname/machine. Also the
clients are a mix of Redhat 6.2/7.2 with 2.2.14-6.0.1/
2.4.9-31 respectively.


-----Original Message-----
From: Benjamin LaHaise [mailto:[email protected]]
Sent: Tuesday, October 08, 2002 12:28 PM
To: [email protected]
Cc: [email protected]
Subject: Re: [NFS] NFS server replacement troubles


On Tue, Oct 08, 2002 at 10:28:52AM -0400, [email protected] wrote:
> Stock Redhat Linux 6.2
> kernel 2.2.14-6.0.1 on i686
> partition being exported is Reiser filesystem

That kernel is utterly ancient. At the very least you should upgrade to the

newer errata that were released for 6.2 to fix security issues in the kernel

and userland, but preferably you should just replace the box with one
running
7.3, as 6.2 is pretty much end of life. Numerous bugs in NFS, reiserfs and
the core kernel have been fixed since 2.2.14 was released almost 2 years
ago.

-ben


-------------------------------------------------------
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

2002-10-09 09:38:42

by NeilBrown

[permalink] [raw]
Subject: Re: NFS server replacement troubles

On Tuesday October 8, [email protected] wrote:
> I currently have a SCO NFS server, however I'm replacing it with Redhat
> Linux.
> I have taking down the box replaced it and NFS fails. The hostname, ip, etc.
> is
> the same on the new machine.
> New box info....
> Stock Redhat Linux 6.2
> kernel 2.2.14-6.0.1 on i686
> partition being exported is Reiser filesystem
>
> Before new box was put in to production its hostname was different.
> Exporting was fine, NFS worked properly, however as soon as box
> is brought up under new name NFS fails, errors(see below ).
>
> Snapshot of errors.......
>
> Starting portmapper: [ OK ]
> master:/var/lib/nfs 23# /etc/rc.d/init.d/nfs start
> Starting NFS quotas: Cannot register service: RPC: Unable to receive; errno
> = Connection refused
> [FAILED]
> Starting NFS mountd: Cannot register service: RPC: Unable to receive; errno
> = Connection refused

At a guess, I would say tcpwrappers problem. Maybe you have to old
host-name in /etc/hosts.allow, or something like that??

NeilBrown

> mountd: unable to register (mountd, 1, udp).
> [FAILED]
> Starting NFS daemon: nfssvc: Connection refused
> [FAILED]
> Starting NFS services: [ OK ]
>
>
> I have restarted portmapper and nfs with no success, unmount nfs on all
> connected machines, still no workee....
> Any help would be appreciated...
> Thanks.
>
>
> Troy L. Swaine
> System Administrator / Engineer
>
>
>
> -------------------------------------------------------
> 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


-------------------------------------------------------
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

2002-10-09 11:20:18

by tswaine

[permalink] [raw]
Subject: RE: NFS server replacement troubles

Thanks Neil but I looked at that also. My exports file looks like
this....

/somedir (rw)

Allowing anyone to connect. Also my hosts.allow is...

ALL: ALL

Allowing everyone also. Right after the new box comes online
and fails(when it worked with a different hostname). Something
gets corrupted in nfs. No matter what I do, nfs will not work.
I have tried replacing binaries ( rpc.* )with no avail.

Could it be stale clients that were not unmounted before the switch?

Thanks....

-----Original Message-----
From: Neil Brown [mailto:[email protected]]
Sent: Wednesday, October 09, 2002 5:38 AM
To: [email protected]
Cc: [email protected]
Subject: Re: [NFS] NFS server replacement troubles


On Tuesday October 8, [email protected] wrote:
> I currently have a SCO NFS server, however I'm replacing it with Redhat
> Linux.
> I have taking down the box replaced it and NFS fails. The hostname, ip,
etc.
> is
> the same on the new machine.
> New box info....
> Stock Redhat Linux 6.2
> kernel 2.2.14-6.0.1 on i686
> partition being exported is Reiser filesystem
>
> Before new box was put in to production its hostname was different.
> Exporting was fine, NFS worked properly, however as soon as box
> is brought up under new name NFS fails, errors(see below ).
>
> Snapshot of errors.......
>
> Starting portmapper: [ OK ]
> master:/var/lib/nfs 23# /etc/rc.d/init.d/nfs start
> Starting NFS quotas: Cannot register service: RPC: Unable to receive;
errno
> = Connection refused
> [FAILED]
> Starting NFS mountd: Cannot register service: RPC: Unable to receive;
errno
> = Connection refused

At a guess, I would say tcpwrappers problem. Maybe you have to old
host-name in /etc/hosts.allow, or something like that??

NeilBrown

> mountd: unable to register (mountd, 1, udp).
> [FAILED]
> Starting NFS daemon: nfssvc: Connection refused
> [FAILED]
> Starting NFS services: [ OK ]
>
>
> I have restarted portmapper and nfs with no success, unmount nfs on all
> connected machines, still no workee....
> Any help would be appreciated...
> Thanks.
>
>
> Troy L. Swaine
> System Administrator / Engineer
>
>
>
> -------------------------------------------------------
> 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


-------------------------------------------------------
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

2002-10-11 06:22:50

by NeilBrown

[permalink] [raw]
Subject: RE: NFS server replacement troubles

On Wednesday October 9, [email protected] wrote:
> Thanks Neil but I looked at that also. My exports file looks like
> this....
>
> /somedir (rw)
>
> Allowing anyone to connect. Also my hosts.allow is...
>
> ALL: ALL
>
> Allowing everyone also. Right after the new box comes online
> and fails(when it worked with a different hostname). Something
> gets corrupted in nfs. No matter what I do, nfs will not work.
> I have tried replacing binaries ( rpc.* )with no avail.
>
> Could it be stale clients that were not unmounted before the switch?

No. It doesn't look at all like a client problem.

The error message all strongly indicate trouble talking to portmap,
which is why I suspected tcpwrappers... Maybe portmap has
died... maybe running strace on portmap will show something...

NeilBrown

>
> Thanks....
>
> -----Original Message-----
> From: Neil Brown [mailto:[email protected]]
> Sent: Wednesday, October 09, 2002 5:38 AM
> To: [email protected]
> Cc: [email protected]
> Subject: Re: [NFS] NFS server replacement troubles
>
>
> On Tuesday October 8, [email protected] wrote:
> > I currently have a SCO NFS server, however I'm replacing it with Redhat
> > Linux.
> > I have taking down the box replaced it and NFS fails. The hostname, ip,
> etc.
> > is
> > the same on the new machine.
> > New box info....
> > Stock Redhat Linux 6.2
> > kernel 2.2.14-6.0.1 on i686
> > partition being exported is Reiser filesystem
> >
> > Before new box was put in to production its hostname was different.
> > Exporting was fine, NFS worked properly, however as soon as box
> > is brought up under new name NFS fails, errors(see below ).
> >
> > Snapshot of errors.......
> >
> > Starting portmapper: [ OK ]
> > master:/var/lib/nfs 23# /etc/rc.d/init.d/nfs start
> > Starting NFS quotas: Cannot register service: RPC: Unable to receive;
> errno
> > = Connection refused
> > [FAILED]
> > Starting NFS mountd: Cannot register service: RPC: Unable to receive;
> errno
> > = Connection refused
>
> At a guess, I would say tcpwrappers problem. Maybe you have to old
> host-name in /etc/hosts.allow, or something like that??
>
> NeilBrown
>
> > mountd: unable to register (mountd, 1, udp).
> > [FAILED]
> > Starting NFS daemon: nfssvc: Connection refused
> > [FAILED]
> > Starting NFS services: [ OK ]
> >
> >
> > I have restarted portmapper and nfs with no success, unmount nfs on all
> > connected machines, still no workee....
> > Any help would be appreciated...
> > Thanks.
> >
> >
> > Troy L. Swaine
> > System Administrator / Engineer
> >
> >
> >
> > -------------------------------------------------------
> > 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


-------------------------------------------------------
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