Skip to content

Latest commit

 

History

History
109 lines (90 loc) · 5.24 KB

FAQ.md

File metadata and controls

109 lines (90 loc) · 5.24 KB
Raw

FAQ

NanoRC exited ungracefully and all my apps are still running, what do I do?

You'll need to go on each of the servers where the applications where running, run htop --user ${USER}, click on the daq_application (yes, click on your terminal) to select the app as shown in the image, then F9 and enter twice.

htop

You can also do the equivalent by doing ps a|grep daq_application, note the first number of the line and run kill <the-number>.

NanoRC can't start the response listener, what do I do?

...i.e., you see an error message which looks like the following:

                    RuntimeError: Cannot create a response listener!!
           ERROR    json0 went to error!                                                     statefulnode.py:225

In this case, you're probably running on the NP04 cluster and you have enabled the web proxy. If so, you can quit nanorc, and run:

source ~np04daq/bin/web_proxy.sh -u

Note that if you do start again right away after sourceing the web proxy file, you may encounter a message such as

Couldn't execute the function boot on the object, reason: The port <hostname>:<portnumber> is already open, likely by another application, cannot continue

This port in question is very likely one which is still open from the connectivity service which was started within your previous nanorc session. At any rate, you can find what that application is (if it's yours) by going to the host in question and running

netstat -tulpn | grep <portnumber>

where the second column will give you the PID of the process. Kill it, and when you've confirmed that it's gone, try again.

NanoRC won't boot my apps?

There are many reasons why this could happen, here are the 2 most common:

  • You don't have password-less ssh keys (see how to create password-less ssh keys)
  • Somebody else is running on the same server, in which case you need to pass --partition-number 1 (or any number between 1 and 10) to the nanorc command.

How do I create password-less ssh keys?

Here is the broad idea (but you are welcome to look on Google too):

ssh-keygen

then tap <Enter> tree times when prompted where to put the key and for a password (do not enter a password here, and the default location for the key is also fine).

Then do:

ssh-copy-id <the-host-where-you-app-will-run>

This command should prompt you a password, for the last time. This is the same password you used to log on the server. After that you can do:

ssh <the-host-where-you-app-will-run>

and you won't be prompted for a password ever again!

I don't want to create ssh keys, how do I do?

This isn't particularly recommended, but if you are running on the dunegpvm you won't have the choice. You can run nanorc with kerberos, by doing nanorc --kerberos ....

Where are my app logs?

NanoRC tells you where the apps logs are:

boot-logs

If you are using with an older version of nanorc, this message won't be here, in this case 2 choices:

  • you are running with nanorc: the logs are in PWD
  • you are running with nano04rc or nanotimingrc: the logs are in /log, and note that these folders are not shared across the NP04 cluster, so you will need to ssh where the app was running first.

I want more logs

Run nanorc like this:

DUNEDAQ_ERS_DEBUG_LEVEL=10 nanorc conf/ partition_name

This will put all the TLOG_DEBUG printout of level <= 10 to print.

I want to send arbitrary data to my application

For that, you can use expert commands. You need to write a json file with the following structure:

{
    "id": "command_id",
    "entry_state": "initial",
    "exit_state": "initial",
    "data": {
        ...some data
    }
}
  • The id entry is the name of the command you have registered on your module (register_command calls for example here).
  • entry and exist state are not necessary (or can be "any"). If you specify them, they need to be the same and nanorc will check that the application is in a correct state.
  • data is the data you are sending to the application. I needs to be keyed with module, etc, just like any configuration data.

Next, after booting your application with nanorc you can do:`

expert_command /config/config/app the-file.json

Where the-file.json is the file you have just created above.and config/config/app is the full path of the application you want to send this command to. So for example if you have started nanorc as such:`

nanorc config session-name

the expert command could be:`

expert_command /config/config/rulocalhosteth0 the-file.json

Ports clashing

The usual answer is that you need to feed a different --partition-number. Nanorc offset automatically ports in your configuration here is what is used:

  • nanorc: port_offset = 0 + partition_number * 500 + subsystem_number * 50
  • nano04rc: port_offset = 0 + partition_number * 500 + subsystem_number * 50
  • nanotimingrc: port_offset = 300 + partition_number * 500 + subsystem_number * 50

If the offsets are the same and you are running nanorc on the same host, you will get port clash errors.