From: Walter Hafner <hafner@informatik.tu-muenchen.de> To: Greg Lehey <grog@lemis.com> Cc: Walter Hafner <hafner@informatik.tu-muenchen.de>, questions@FreeBSD.ORG Subject: Re: strange NFS problem Message-ID: <14218.63691.589830.15439@hprbg5.informatik.tu-muenchen.de> In-Reply-To: <19990713112452.U21403@freebie.lemis.com> References: <14218.4014.47678.998590@hprbg5.informatik.tu-muenchen.de> <srjwvw5dd71.fsf@hprbg5.informatik.tu-muenchen.de> <19990713112452.U21403@freebie.lemis.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Thanks for your help. Here's a little more info to clarify things:
> On Monday, 12 July 1999 at 17:54:22 +0200, Walter Hafner wrote:
> > Host A: FreeBSD 2.2.8 STABLE, NFS Server
> > Host B: FreeBSD 3.1 RELEASE, NFS Client
> >
> > Host A runs "mountd -n" and "nfsd" (4 childs).
> >
> > Routing between A and B is ok, DNS and reverse DNS is ok.
> >
> > /etc/exports on A:
> >
> > /export/dist B C D
> > /export/data B C D
> >
> > /etc/fstab on B:
> >
> > A:/export/data /mount/A/export/data nfs 2,rw,intr,soft,noauto 0 0
> > A:/export/dist /mount/A/export/dist nfs 2,rw,intr,soft,noauto 0 0
> >
> > When I do a
> >
> > mount /mount/w3proj1/export/data
> >
> > on B it hangs forever. When I start mountd with -d on A, it tells me
> > "mount successful" or something similar (don't have the machine at
> > hand.)
> >
> > A ktrace on the "mount ..." line gives:
> >
> > [std lib messages]
> > [reading fstab]
> > 53205 mount RET read 528/0x210
> > 53205 mount CALL statfs(0x806bfbd,0xefbfd0b8)
> > 53205 mount NAMI "/"
> > 53205 mount RET statfs 0
> > 53205 mount CALL stat(0x8071320,0xefbfd1b8)
> > 53205 mount NAMI "/dev/da0s1a"
> > 53205 mount RET stat 0
> > 53205 mount CALL open(0x806ba40,0,0xefbfcb3c)
> > 53205 mount NAMI "."
> > 53205 mount RET open 4
> > 53205 mount CALL chdir(0xefbfcb3c)
> > 53205 mount NAMI "/mount/A/export"
> > 53205 mount RET chdir 0
> > 53205 mount CALL lstat(0xefbfcb52,0xefbfcaac)
> > 53205 mount NAMI "data"
> > 53205 mount PSIG SIGINT SIG_DFL
>
> Is system B multi-homed? In other words, does it have more than one
> IP address? There's a bug in the system which will cause it to accept
> the mount, but possibly reply with the "other" IP address. This will
> cause a hang which you can only resolve with a reboot. See PR
> kern/9612 for more details.
"A", the NFS _server_ is multi-homed. It answers to 6 IP-addresses and
about 15 names. It runs 3 IP-based virtual WWW-servers and about 10
name-based virtual WWW-servers (HTTP 1.1; one IP address). The rest of
the IP numbers and names are assigned for special services (mail, MySQL,
DNS, ...).
I know, what you want to say now ("NFS is Bad(tm) for WWW machines"),
but all the servers are pretty low volume, so there's no I/O bandwidth
problem. It's just, that editing pages is much more convenient via NFS.
I mounted from the DNS-Server CNAME, not from the primary name.
Machine "B" is an ordinary workstation (one IP-address, one name)
> > Never mind. I found the problem (apart from the bogus "2" option in
> > /etc/fstab :-)
> >
> > I didn't mount from the machines "A" record name, but from on of the
> > machines "CNAME"s. The FreeBSD 3.1 mount doesn't seem to like this.
>
> If my reasoning above is correct, I don't think that's the solution,
> unless by changing the name you removed the duplicate IP addresses.
I don't quite understand, what you mean by "removed the duplicate IP
addresses". All addresses are still valid. I'm no kernel hacker. All I
can say is: it works now. I tried both configurations (mountpoints) and
can repeat the behaviour at any time.
Looks like the behaviour you describe in 9612. I can especcially verify
this part:
: Here the mount succeeds on the server side, but the reply contains the
: address 192.109.197.137, not 139.130.136.133, and is thus rejected by
: panic. The mount process hangs in sbwait and is not stoppable.
But it seems to me, that mounts from the name in the "A" record always
succeed.
If I can be any help in tracking the bug down, I'll help. However, the
machine is a production machine and I can't reboot every few minutes
just to debug the kernel.
Regards,
-Walter
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?14218.63691.589830.15439>