sisuite-users

[sisuite-users] rsync issue with image retrieval from golden client

[Doug H. <dou...@va...>]

Greetings, SystemImager persons.  I have an issue for which I must turn to the 
experts.  I've used SystemImager successfully before (on a different set of 
machines than described here), so I'm not a _total_ novice, but I'm not an 
expert user, either.

I have a small cluster of Ubuntu 8.04 nodes onto which I have installed 
systemimager 4.02 and systemconfigurator 2.2.11 from the Debian packages on the 
website.  On my golden client (called 'node02') I can create an image with 
si_prepareclient, no problem (run as root).  However, attempting to retrieve the 
image from the image server (the cluster's master node) results in the following:

<transcript>
# si_getimage --version
si_getimage (part of SystemImager) v4.0.2
# si_getimage --golden-client node02 --image image_name --ip-assignment dhcp 
--post-install kexec
[snip...  saying 'y' to questions]

Retrieving /etc/systemimager/mounted_filesystems from node02 to check for 
mounted filesystems...
------------- node02 mounted_filesystems RETRIEVAL PROGRESS -------------
@ERROR: access denied to root from node01 (192.168.44.1)
rsync error: error starting client-server protocol (code 5) at main.c(1383) 
[receiver=2.6.9]
------------- node02 mounted_filesystems RETRIEVAL FINISHED -------------
Failed to retrieve /etc/systemimager/mounted_filesystems from node02.
si_getimage: Have you run "si_prepareclient" on node02?
           If you see the message "unrecognised option" above, check
           http://systemimager.org/download/ to be sure that you are running
           the recommended version of rsync.
#
</transcript>

And yet, rsync between the two machines, for that particular file even, WILL 
work, as you can see in this transcript:

<transcript>
# mkdir /tmp/foo
# cd /tmp/foo
# rsync --version
rsync  version 2.6.9  protocol version 29
# FILE=/etc/systemimager/mounted_filesystems
# rsync --log-file=test_rsync_log.txt  root@node02:$FILE .
# ls
total 8
-rw-r--r-- 1 root root 662 2008-12-01 09:46 mounted_filesystems
-rw-r--r-- 1 root root 182 2008-12-01 09:46 test_rsync_log.txt
#  wtf!
-bash: wtf!: command not found
#
</transcript>

There you see it, rsync got the file from node02 with nary a problem.  I'm 
pretty sure it's not a firewall issue, since there are no iptables rules in 
effect on the golden client, and even if there were, you can see that the simple 
rsync example worked (there is a firewall active on the image server, but the 
rsync port 873 is open, and again, the simple example works).

Is there something simple I'm missing?  A systemimager config file somewhere?  I 
don't remember having to do such a thing from my previous experience.

I have Googled the error message thoroughly, including permutations, to no 
avail.  Frustratingly, I can't even seem to find out exactly what "code 5" is.

Please let me know if you need more information on context...  I'd be most 
grateful for any help that you could offer.

Many thanks,
Doug

Re: [sisuite-users] rsync issue with image retrieval from golden client

[Doug H. <dou...@va...>]

I thought I'd try this once more...  Anyone have any thoughts on the issue below?

Thanks,
Doug




Doug Hackworth wrote:
> Greetings, SystemImager persons.  I have an issue for which I must turn to the 
> experts.  I've used SystemImager successfully before (on a different set of 
> machines than described here), so I'm not a _total_ novice, but I'm not an 
> expert user, either.
> 
> I have a small cluster of Ubuntu 8.04 nodes onto which I have installed 
> systemimager 4.02 and systemconfigurator 2.2.11 from the Debian packages on the 
> website.  On my golden client (called 'node02') I can create an image with 
> si_prepareclient, no problem (run as root).  However, attempting to retrieve the 
> image from the image server (the cluster's master node) results in the following:
> 
> <transcript>
> # si_getimage --version
> si_getimage (part of SystemImager) v4.0.2
> # si_getimage --golden-client node02 --image image_name --ip-assignment dhcp 
> --post-install kexec
> [snip...  saying 'y' to questions]
> 
> Retrieving /etc/systemimager/mounted_filesystems from node02 to check for 
> mounted filesystems...
> ------------- node02 mounted_filesystems RETRIEVAL PROGRESS -------------
> @ERROR: access denied to root from node01 (192.168.44.1)
> rsync error: error starting client-server protocol (code 5) at main.c(1383) 
> [receiver=2.6.9]
> ------------- node02 mounted_filesystems RETRIEVAL FINISHED -------------
> Failed to retrieve /etc/systemimager/mounted_filesystems from node02.
> si_getimage: Have you run "si_prepareclient" on node02?
>            If you see the message "unrecognised option" above, check
>            http://systemimager.org/download/ to be sure that you are running
>            the recommended version of rsync.
> #
> </transcript>
> 
> And yet, rsync between the two machines, for that particular file even, WILL 
> work, as you can see in this transcript:
> 
> <transcript>
> # mkdir /tmp/foo
> # cd /tmp/foo
> # rsync --version
> rsync  version 2.6.9  protocol version 29
> # FILE=/etc/systemimager/mounted_filesystems
> # rsync --log-file=test_rsync_log.txt  root@node02:$FILE .
> # ls
> total 8
> -rw-r--r-- 1 root root 662 2008-12-01 09:46 mounted_filesystems
> -rw-r--r-- 1 root root 182 2008-12-01 09:46 test_rsync_log.txt
> #  wtf!
> -bash: wtf!: command not found
> #
> </transcript>
> 
> There you see it, rsync got the file from node02 with nary a problem.  I'm 
> pretty sure it's not a firewall issue, since there are no iptables rules in 
> effect on the golden client, and even if there were, you can see that the simple 
> rsync example worked (there is a firewall active on the image server, but the 
> rsync port 873 is open, and again, the simple example works).
> 
> Is there something simple I'm missing?  A systemimager config file somewhere?  I 
> don't remember having to do such a thing from my previous experience.
> 
> I have Googled the error message thoroughly, including permutations, to no 
> avail.  Frustratingly, I can't even seem to find out exactly what "code 5" is.
> 
> Please let me know if you need more information on context...  I'd be most 
> grateful for any help that you could offer.
> 
> Many thanks,
> Doug
> 
> 

Re: [sisuite-users] rsync issue with image retrieval from golden client

[J. C. <jc...@mo...>]

Doug Hackworth <doug.hackworth@...> writes:
> 
> > [snip...  saying 'y' to questions]
> > 
> > Retrieving /etc/systemimager/mounted_filesystems from node02 to check for 
> > mounted filesystems...
> > ------------- node02 mounted_filesystems RETRIEVAL PROGRESS -------------
> > @ERROR: access denied to root from node01 (192.168.44.1)
> > rsync error: error starting client-server protocol (code 5) at main.c(1383) 
> > [receiver=2.6.9]
> > ------------- node02 mounted_filesystems RETRIEVAL FINISHED -------------
> > Failed to retrieve /etc/systemimager/mounted_filesystems from node02.
> > si_getimage: Have you run "si_prepareclient" on node02?
> >            If you see the message "unrecognised option" above, check
> >            http://systemimager.org/download/ to be sure that you are running
> >            the recommended version of rsync.
> > #
> > </transcript>
> > 


I ran into this issue tonight with the same error messages from si_getclient
running on the server. But for me the problem was simply that the --server ip
option for si_prepareclient on the golden client did not match the ip of the
server that I was running si_getclient on to attempt to retrieve the golden
image.

On my golden client, I was running this manually:

   $ si_prepareclient --server 10.1.12.238 -e /dev/sda --no-uyok

But my server's ip is 10.1.12.244, so obviously, I was getting this error from
si_getclient.

Re: [sisuite-users] rsync issue with image retrieval from golden client

[Bernard L. <be...@va...>]

Hi Doug:

On Mon, Dec 1, 2008 at 10:19 AM, Doug Hackworth
<dou...@va...> wrote:

> ------------- node02 mounted_filesystems RETRIEVAL PROGRESS -------------
> @ERROR: access denied to root from node01 (192.168.44.1)

Perhaps you have permission problems grabbing files from node02's
/root from node01?

> # FILE=/etc/systemimager/mounted_filesystems
> # rsync --log-file=test_rsync_log.txt  root@node02:$FILE .
> # ls
> total 8
> -rw-r--r-- 1 root root 662 2008-12-01 09:46 mounted_filesystems
> -rw-r--r-- 1 root root 182 2008-12-01 09:46 test_rsync_log.txt
> #  wtf!
> -bash: wtf!: command not found

This isn't what si_getimage is doing though:

http://trac.systemimager.org/browser/trunk/sbin/si_getimage

341 	    $command = "rsync -av --numeric-ids";
342 	    if ($log) { $command = $command . qq( --log-format="$log"); }
343 	    if ($ssh_user) { $command = $command . " --bwlimit=10000"; }
344 	    $command =  $command . "
rsync://${source_host}:${port}/root/etc/systemimager/mounted_filesystems
${imagedir}/etc/systemimager/mounted_filesystems";

Try this command instead -- note it is trying to retrieve from
/root/etc/... not /etc/... from the ${source_host}.

Cheers,

Bernard

Re: [sisuite-users] rsync issue with image retrieval from golden client

[Doug H. <dou...@va...>]

Hi, Bernard.  Many thanks for pointing out the precise command that's being 
executed -- that's a most helpful insight.  I actually contrived a workaround 
for my problem several days ago, but I'm keen to know what was causing it.


>> ------------- node02 mounted_filesystems RETRIEVAL PROGRESS -------------
>> @ERROR: access denied to root from node01 (192.168.44.1)
> 
> Perhaps you have permission problems grabbing files from node02's
> /root from node01?

I thought of that, but it appears to have not been the problem -- I could rsync 
(and scp) files located in node02:/root from node01:/root with no trouble.



>> # FILE=/etc/systemimager/mounted_filesystems
>> # rsync --log-file=test_rsync_log.txt  root@node02:$FILE .
>
> This isn't what si_getimage is doing though:
> 
> http://trac.systemimager.org/browser/trunk/sbin/si_getimage
> 
> 341 	    $command = "rsync -av --numeric-ids";
> 342 	    if ($log) { $command = $command . qq( --log-format="$log"); }
> 343 	    if ($ssh_user) { $command = $command . " --bwlimit=10000"; }
> 344 	    $command =  $command . "
> rsync://${source_host}:${port}/root/etc/systemimager/mounted_filesystems
> ${imagedir}/etc/systemimager/mounted_filesystems";
> 
> Try this command instead -- note it is trying to retrieve from
> /root/etc/... not /etc/... from the ${source_host}.

When I pieced together the command from the code you quoted and ran it -- no 
problem, it ran successfully.  Encouraged, I decided to try si_getimage again... 
and it worked, too, even though it "shouldn't" have.  So something I did between 
now and the last time I attempted to run si_getimage magically cured the 
problem, a situation both pleasantly surprising and thorougly maddening.  I 
genuinely am not sure what it might have been since I didn't do anything "major" 
other than a few reboots.  Ah, well... For now I'll just enjoy the fact that 
it's working.  If I make a further discovery later I'll post it here.

Thanks,
Doug