[prev in list] [next in list] [prev in thread] [next in thread] 

List:       npaci-rocks-discussion
Subject:    [Rocks-Discuss]   : Ganglia not seeing nodes
From:       "Werf, C.G. van der (Carel)" <C.G.vanderWerf () uu ! nl>
Date:       2015-06-26 8:38:23
Message-ID: 458F5B82F22A4C4AA560DA0D896D57C94A98657D () WP0047 ! soliscom ! uu ! nl
[Download RAW message or body]

Hi,
I think I finally figured out why on my new Rocks 6.2 installm Ganglia will not see \
the nodes.

This might actually be a Rocks 6.2 bug.

On my frontnode (phobos):

# rocks list host route phobos

HOST    NETWORK         NETMASK         GATEWAY        SOURCE
phobos: 0.0.0.0         0.0.0.0         131.211.32.129 H     
phobos: 131.211.32.171  255.255.255.255 192.168.40.100 G     
phobos: 224.0.0.0       255.255.255.0   em2            G     
phobos: 255.255.255.255 255.255.255.255 em2            G     

but:
# route -n

Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
131.211.32.171  192.168.40.100  255.255.255.255 UGH   0      0        0 em2
131.211.32.128  0.0.0.0         255.255.255.128 U     0      0        0 em1
192.168.40.0    0.0.0.0         255.255.255.0   U     0      0        0 em2
169.254.0.0     0.0.0.0         255.255.0.0     U     1002   0        0 em1
169.254.0.0     0.0.0.0         255.255.0.0     U     1003   0        0 em2
0.0.0.0         131.211.32.129  0.0.0.0         UG    0      0        0 em1

So there the 224.0.0.0 route is missing !

"rocks sync host network" doesn't fix this.

Any clue, why the rocks command misses this route ?

Regards,
Carel

-----Original Message-----
From: npaci-rocks-discussion-bounces+c.g.vanderwerf=uu.nl@sdsc.edu \
[mailto:npaci-rocks-discussion-bounces+c.g.vanderwerf=uu.nl@sdsc.edu] On Behalf Of \
                Werf, C.G. van der (Carel)
Sent: 24 June, 2015 15:23
To: 'Discussion of Rocks Clusters'
Subject: [Rocks-Discuss] Re: unable to pxe-boot

Hi,

I've previously installed a Rocks-6.1.1 cluster which comletely works as it should.

Now, I recently mentioned on this list about network-problems with a new cluster, \
caused by the reason that I had to change the public and private Ips of the headnode. \
At the end I fixed all anomalies, such as pxe-boot-server address and routing tables, \
but I was unable to get ganglia see the compute node (I just started with 1 compute \
node...). Then I decided to start allover and reinstall the frontnode with Rocks-6.2, \
and deploy 1 compute node. All seems fine now, but I still cannot get ganglia to see \
more than the frontnode.

What is missing here ?

- /etc/ganglia/gmond.conf seems right on both frontnode and compute node
- /etc/ganglia/gmetad.conf seems right on frontnode
- services gmetad and gmond restarted on frontnode, gmond restarted on compute-node.

Still no luck.

- $ gstat -a only shows frontnode.

Private network is on a basic DELL switch.

How can I test 'multi-cast'-ability of my private network ?

Regards,
Carel 


[prev in list] [next in list] [prev in thread] [next in thread]