[Rocks-Discuss] rocks list host profile compute-0-0 traceback syntax error

441 views
Skip to first unread message

Sarlo, Jeffrey S

unread,
Mar 15, 2012, 10:42:28 AM3/15/12
to npaci-rocks...@sdsc.edu
There are some errors when trying to install a node and the following error from rocks list host profile is probably causing it. We have run xmllint against our extend-compute.xml and replace-partition.xml and there were no errors. Any ideas on how to fix this would be appreciated.

# rocks list host profile compute-0-0

Traceback (most recent call last):
File "/opt/rocks/bin/rocks", line 294, in ?
command.runWrapper(name, args[i:])
File "/opt/rocks/lib/python2.4/site-packages/rocks/commands/__init__.py", line 2120, in runWrapper
self.run(self._params, self._args)
File "/opt/rocks/lib/python2.4/site-packages/rocks/commands/list/host/profile/__init__.py", line 298, in run
[
File "/opt/rocks/lib/python2.4/site-packages/rocks/commands/__init__.py", line 1802, in command
o.runWrapper(name, args)
File "/opt/rocks/lib/python2.4/site-packages/rocks/commands/__init__.py", line 2120, in runWrapper
self.run(self._params, self._args)
File "/opt/rocks/lib/python2.4/site-packages/rocks/commands/list/host/xml/__init__.py", line 202, in run
xml = self.command('list.node.xml', args)
File "/opt/rocks/lib/python2.4/site-packages/rocks/commands/__init__.py", line 1802, in command
o.runWrapper(name, args)
File "/opt/rocks/lib/python2.4/site-packages/rocks/commands/__init__.py", line 2120, in runWrapper
self.run(self._params, self._args)
File "/opt/rocks/lib/python2.4/site-packages/rocks/commands/list/node/xml/__init__.py", line 520, in run
handler.parseNode(node, doEval)
File "/opt/rocks/lib/python2.4/site-packages/rocks/profile.py", line 391, in parseNode
parser.feed(handler.getXMLHeader())
File "/opt/rocks/lib/python2.4/site-packages/_xmlplus/sax/expatreader.py", line 220, in feed
self._err_handler.fatalError(exc)
File "/opt/rocks/lib/python2.4/site-packages/_xmlplus/sax/handler.py", line 38, in fatalError
raise exception
xml.sax._exceptions.SAXParseException: <unknown>:66:20: syntax error

Thanks.
Jeff

Philip Papadopoulos

unread,
Mar 15, 2012, 11:27:39 AM3/15/12
to Discussion of Rocks Clusters
If you remove your extend-compute.xml and/or replace-partition.xml and
rebuild the distro, does the problem go away?
if yes, then send your extend-compute.xml and/or replace-partition.xml file
and we'll take a look at it.

-P


--
Philip Papadopoulos, PhD
University of California, San Diego
858-822-3628 (Ofc)
619-331-2990 (Fax)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://lists.sdsc.edu/pipermail/npaci-rocks-discussion/attachments/20120315/45177c5e/attachment.html

Greg Bruno

unread,
Mar 15, 2012, 11:31:25 AM3/15/12
to Discussion of Rocks Clusters
On Thu, Mar 15, 2012 at 7:42 AM, Sarlo, Jeffrey S <JSa...@central.uh.edu> wrote:
> There are some errors when trying to install a node and the following error from rocks list host profile is probably causing it.  We have run xmllint against our extend-compute.xml and replace-partition.xml and there were no errors.  Any ideas on how to fix this would be appreciated.

Try:

# ROCKSDEBUG=y rocks list host profile compute-0-0

Then look at the last couple lines -- it should be obvious in which
node XML file and which line the syntax error is on.

- gb

Sarlo, Jeffrey S

unread,
Mar 15, 2012, 12:27:34 PM3/15/12
to npaci-rocks...@sdsc.edu
Removing the extend-compute.xml and replace-partition.xml and rebuilding ends up with the same messages.

Same results as before with ROCKSDEBUG.
Forgot to put that this is ROCKS 5.4.3.

# ROCKSDEBUG=y rocks list host profile compute-0-0

Sarlo, Jeffrey S

unread,
Mar 15, 2012, 2:17:28 PM3/15/12
to npaci-rocks...@sdsc.edu

We found the problem. There was an attribute in the database for each node that had a % in it. We removed that attribute and now everything is working.

Thanks.
Jeff

Riley, Michael F. (LARC-D309)[LITES]

unread,
Mar 16, 2012, 10:26:29 AM3/16/12
to Discussion of Rocks Clusters
Does anyone have additional documentation for insert-ethers. The --help only reveals the potential of the command.

./sbin/insert-ethers --help
Insert Ethernet Addresses - version 5.4
Usage: insert-ethers [-hv] [-p password] [-u host] [-d database] [--help]
[--list-rcfiles] [--list-project-info] [--verbose] [--update] [--staticip]
[--rcfile arg] [--host host] [--password password] [--db database]
[--user host] [--baseip ip address] [--basename basename] [--hostname hostname]
[--ipaddr ip address] [--cabinet number] [--rack number] [--inc number]
[--rank number] [--replace hostname] [--remove hostname]
[--os the OS to install on the machines]


Thanks
Mike Riley
m.f....@nasa.gov

Philip Papadopoulos

unread,
Mar 16, 2012, 2:01:53 PM3/16/12
to Discussion of Rocks Clusters
What do you want to do? Then I can help guide you through the menagerie of
flags :-(.

-P

--
Philip Papadopoulos, PhD
University of California, San Diego
858-822-3628 (Ofc)
619-331-2990 (Fax)
-------------- next part --------------
An HTML attachment was scrubbed...

URL: https://lists.sdsc.edu/pipermail/npaci-rocks-discussion/attachments/20120316/c70d70cb/attachment.html

Riley, Michael F. (LARC-D309)[LITES]

unread,
Mar 16, 2012, 2:44:48 PM3/16/12
to Discussion of Rocks Clusters
We want to add IPMI interfaces /hostnames that match the compute node number.

compute-0-59 would relate to power-0-59

Since we are adding new nodes, we would like to insert-ethers to start at the same node number as the compute node and continue on. Since we have no power-0-xx names, insert-ethers wants to start at power-0-0 and our compute nodes are at 0-59 . I can do them one at a time by insert-ethers -- hostname power-0-59 , but it is time consuming for a whole rack of nodes.

I was hoping to start insert-ethers and then plug in the network inferface one at time while they are discovered and assigned a name.

Thanks
Mike

________________________________________
From: npaci-rocks-dis...@sdsc.edu [npaci-rocks-dis...@sdsc.edu] On Behalf Of Philip Papadopoulos [philip.pa...@gmail.com]
Sent: Friday, March 16, 2012 2:01 PM
To: Discussion of Rocks Clusters
Subject: Re: [Rocks-Discuss] insert-ethers documentation?

Riley, Michael F. (LARC-D309)[LITES]

unread,
Mar 16, 2012, 3:15:25 PM3/16/12
to Discussion of Rocks Clusters

Philip Papadopoulos

unread,
Mar 16, 2012, 3:27:18 PM3/16/12
to Discussion of Rocks Clusters
We handle this a bit differently by making the IPMI interface a host
interface.
Do the following ONCE
# rocks add network ipmi <subnet> netmask=<netmask>

for each host
# rocks add host interface <host> ipmi subnet=ipmi ip=<ip address>

Then
# rocks sync config
# rocks sync host network compute


If you look at the python script at
http://git.rocksclusters.org/cgi-bin/gitweb.cgi?p=triton/scar/.git;a=blob;f=src/root-scripts/sbin/addInterface;h=d07652082875e986816467e73f511773d7f7116c;hb=e2497feb8e350a28042436408fb95ea974c8cb40

this is a local utility we wrote to match the last octets so that different
subnets on the same host have the
same last octets.

On Fri, Mar 16, 2012 at 11:44 AM, Riley, Michael F. (LARC-D309)[LITES] <
m.f....@nasa.gov> wrote:

> We want to add IPMI interfaces /hostnames that match the compute node
> number.
>
> compute-0-59 would relate to power-0-59
>
> Since we are adding new nodes, we would like to insert-ethers to start at
> the same node number as the compute node and continue on. Since we have
> no power-0-xx names, insert-ethers wants to start at power-0-0 and our
> compute nodes are at 0-59 . I can do them one at a time by insert-ethers
> -- hostname power-0-59 , but it is time consuming for a whole rack of nodes.
>
> I was hoping to start insert-ethers and then plug in the network
> inferface one at time while they are discovered and assigned a name.
>

If you want to go the insert ethers way, then use the rank option
eg insert-ethers --rank=59
to begin at rank 59.
(please test that we assign the rank first and then increment, rather than
the other way around)

-P

URL: https://lists.sdsc.edu/pipermail/npaci-rocks-discussion/attachments/20120316/c12aec97/attachment.html

Reply all
Reply to author
Forward
0 new messages