NodeHandler FAQ

Here you can find the answer to some of the most frequently asked questions about NodeHandler. If you have a question not answered by this page you can ask it on the MailingList.

Questions answered in seperate pagess:

• How to control nodes?
• Which Tools are needed to create a debian package?
• How to Use Noise Generator in experiments?
• Sandbox IP convention?
• How to Program a Node ID box?

How to check the status of my experiment?

Point your browser to ​http://remote.orbit-lab.org:4000/xml

How do I log into a node to see what is happening?

From remote.orbit-lab.org, you can use the serial console access provided by CM as follows

telnet 10.1.x.y 3025

How do I check the status of the imaging process?

Point your browser to ​http://remote.orbit-lab.org:4000/progress

How to interpret the progress bar?

INFO exp: Progress: 30/45/60 min/avg/max (84.571326) For the imaging process, this means that slowest node is 30% done and fastest node is 60% done The average might require an additional explanation.

INFO exp: Progress: 90/85/100 min/avg/max (234.240859)

The average is calculated as the progress over ALL nodes, while the min/max are for the nodes which have started frisbee already. In the above case, a few nodes hadn't booted up yet.

Where are the noise antennas located?

• Antenna 1: Between node2-1 and node2-2
• Antenna 2: Between node2-7 and node2-8
• Antenna 3: Between node7-1 and node7-2
• Antenna 4: Between node7-7 and node7-8.

[Kishore]

Where can I find information about the wireless cards used in Orbit

Please look at the madwifi driver for atheros cards, the ipw2200 driver for Intel cards and related documentation online. I believe the model no. for atheros cards is AR5212. [Kishore]

Which interface is which?

For Atheros based nodes, ath1 is near the power supply, ath0 is near the serial port For Intel based nodes, eth3 is near the power supply, eth2 is near the serial port

It seems that Node[5,6] cannot receive any packet from Node[7,4]

If you assign the netmask in the script to be 255.255.255.0 it filters out all nodes but something1.something2.somethin3.x. If you assign 255.255.0.0 all the nodes on the grid are on the same subnet. [Haris]

When I run 'ntpdate ntpservername' on a node, I got an error that the ntp port is used already.

The nodes run ntpd as a service which occupies the port and hence you get the error when you try to manually run ntpdate. You could disable ntpd, run ntpdate manually and then enable ntpd again to overcome the slight drifts. [Sachin]

Interpreting error messages

Error message

FATAL run: Exception: Can't switch off node ALL:ALL (#<Net::HTTPServerError:0xb7be7ce4>) RuntimeError

/tmp/eee.972/lib/handler/cmc.rb:46:in `nodeOff'

Fix: CMC service is not running (or dead), need to restart CMC service

Where can I find out more about the schema used in the Orbit database?

Is this some thing I design myself for my own purposes or is there some previously designed schema that everybody uses? How does one experiment code write stuff to the database, or does this happen automatically?

ORBIT Services will create ONE database (same name as experiment ID) per experiment. However, the database may contain multiple tables. One for each application/configuration used in the experiment and - soon, but not yet - an experiment specific table containing all the experiment related information.

Each application (in the definition file) tells nodehandler what is needs as inputs and what are the statistics it will report. Accordingly nodehandler instructs the oml grid service to create a table in the database with the appropriate schema.

The database schema for the receiver side information is as follows Where rcvd_pkt_size_sample_sum = throughput

Field

Type

node_id	varchar(32)
sequence_no	int(11)
timestamp	int(11)
stream_no	int(11)
pkt_seqno	int(11)
sender_port	int(11)
flow_no	int(11)
pkt_num_rcvd	int(11)
rcvd_pkt_size_sample_sum	int(11)
rx_timestamp	int(11)
rssi	int(11)
xmitrate_sample_mean	float