Skip to contentSkip to conversation

Failure to mount without ever trying to mount..

DISCUSSION Michael Marion 3 participants 4 messages May 31, 2005 – Jun 1, 2005 Kernel automounter

lore
Share & tools
lore
#0 lore raw

Wonder if this is known.. don't see the same thing in the list, but I'm way behind.

Most hosts failing to mount a test server, but watching packets shows that it never does try to mount. I've verified that patching 4.1.4 with:
autofs-4.1.4-multi-parse-fix.patch autofs-4.1.4-non-replicated-ping.patch autofs-4.1.4-misc-fixes.patch

fixes the problem, and I'm guessing it's the non-replicated-ping patch that does it as the ok hosts do a lot more rpc traffic.

Only this one host fails, and I think it's due to it not running nfs v2 or something else under portmap, it's only running:
cornholio src {511}$ rpcinfo -p box
program vers proto port

    100000    2   udp    111  portmapper
    100000    2   tcp    111  portmapper
    100003    3   udp   2049  nfs
    100003    3   tcp   2049  nfs
    100005    1   udp    635  mountd
    100005    1   tcp    635  mountd
    100005    2   udp    635  mountd
    100005    2   tcp    635  mountd
    100005    3   udp    635  mountd
    100005    3   tcp    635  mountd

Most servers (and filers) which work fine have more ports up, like nfs v2, status, nlockmgr, etc.

We're trying to figure out if just enabling nfs v2 support, or something else simple will help in the short term.. I'd rather not update the kernel and automount daemon on 800+ hosts for one path. :) Though I want to get them all updated eventually where I can.

I attached ethereal caps of a successful host vs a failed host... the failed host does getport call for nfs v3, talks to 2049, does an nfs v3 null call.. more nfsd port talk, getport call for nfs v2, then quits.

Signature · 3 lines
Mike Marion-Unix SysAdmin/Staff Engineer-http://www.qualcomm.com
Stewie Griffin: Forecast for tomorrow; A few sprinkles of genius with a chance of 
doom.  ==> Family Guy
#1 lore raw
Quoted 11 lines · Michael Marion

Wonder if this is known.. don't see the same thing in the list, but I'm way behind.

Most hosts failing to mount a test server, but watching packets shows that it never does try to mount. I've verified that patching 4.1.4 with:
autofs-4.1.4-multi-parse-fix.patch autofs-4.1.4-non-replicated-ping.patch autofs-4.1.4-misc-fixes.patch

fixes the problem, and I'm guessing it's the non-replicated-ping patch that does it as the ok hosts do a lot more rpc traffic.

If patching autofs works, why not simply do that?

Quoted 4 lines

We're trying to figure out if just enabling nfs v2 support, or something else simple will help in the short term.. I'd rather not update the kernel and automount daemon on 800+ hosts for one path. :) Though I want to get them all updated eventually where I can.

You don't have to update the kernel for the fixes, just the daemon. If you have 800+ machines, surely you must have some kind of automated system for doing that.

/* Steinar */

Signature · 1 line
Homepage: http://www.sesse.net/
maintainer
#2 lore raw
Quoted 15 lines · Steinar H. Gunderson
Quoted 11 lines · Michael Marion

Wonder if this is known.. don't see the same thing in the list, but I'm way behind.

Most hosts failing to mount a test server, but watching packets shows that it never does try to mount. I've verified that patching 4.1.4 with:
autofs-4.1.4-multi-parse-fix.patch autofs-4.1.4-non-replicated-ping.patch autofs-4.1.4-misc-fixes.patch

fixes the problem, and I'm guessing it's the non-replicated-ping patch that does it as the ok hosts do a lot more rpc traffic.

If patching autofs works, why not simply do that?

Quoted 8 lines
Quoted 4 lines

We're trying to figure out if just enabling nfs v2 support, or something else simple will help in the short term.. I'd rather not update the kernel and automount daemon on 800+ hosts for one path. :) Though I want to get them all updated eventually where I can.

You don't have to update the kernel for the fixes, just the daemon. If you have 800+ machines, surely you must have some kind of automated system for doing that.

Yep. Assuming I'm right. Resolving this specific issue won't need a kernel update.

Ian

maintainer
#3 lore raw
In reply to Michael Marion
Quoted 11 lines · Michael Marion

Wonder if this is known.. don't see the same thing in the list, but I'm way behind.

Most hosts failing to mount a test server, but watching packets shows that it never does try to mount. I've verified that patching 4.1.4 with:
autofs-4.1.4-multi-parse-fix.patch autofs-4.1.4-non-replicated-ping.patch autofs-4.1.4-misc-fixes.patch

fixes the problem, and I'm guessing it's the non-replicated-ping patch that does it as the ok hosts do a lot more rpc traffic.

Exactly right.
The problem was in the probe logic. It fails if a machine doesn't respond quickly enough rather than retrying with a longer timeout.

Maybe this one machine has some other problem that's slowing it down? Duplex?

Quoted 16 lines

Only this one host fails, and I think it's due to it not running nfs v2 or something else under portmap, it's only running:
cornholio src {511}$ rpcinfo -p box
program vers proto port

    100000    2   udp    111  portmapper
    100000    2   tcp    111  portmapper
    100003    3   udp   2049  nfs
    100003    3   tcp   2049  nfs
    100005    1   udp    635  mountd
    100005    1   tcp    635  mountd
    100005    2   udp    635  mountd
    100005    2   tcp    635  mountd
    100005    3   udp    635  mountd
    100005    3   tcp    635  mountd
Quoted 7 lines

Most servers (and filers) which work fine have more ports up, like nfs v2, status, nlockmgr, etc.

We're trying to figure out if just enabling nfs v2 support, or something else simple will help in the short term.. I'd rather not update the kernel and automount daemon on 800+ hosts for one path. :) Though I want to get them all updated eventually where I can.

The only thing to do is make the machine respond more quickly. Network card driver update?

Ian