question

[Xin Yao]

Dear Powder Team,

I have tried the modified profile based on the srslte-otalab profile and fixed several problems to make it work at first in today's experiment, but then I found some problems when I logged in to the node using the ssh command.

1. I found that for both X310 and B210, the pre-setting files(".sh" in the /bin folder) are not completely processed. After I checked the parameters on the X310 and B210 node using the " sysctl -a" command, it seems that only the "/bin/add-nat-and-ip-forward.sh" file has been processed.

2. I also found the srsran has yet to be updated from srslte since the "/bin/deploy.sh" file was not processed.

3. Moreover, after all the pre-setting is successfully processed since I add a computing node in this profile, I wonder if there is any way I can let the additional computing node do the computation process (in code) for the node running the srsran protocol (the additional computing node is connected to the existing node by LAN).

I am sorry that I cannot provide the experiment link since I have already terminated the experiment.

Here is the link to the repository for the modified profile: https://github.com/XYao2021/srsran_modified.git

Best wishes,

Xin Yao

[Kirk Webb]

Xin,

During our most recent office hours meeting, I pointed you at an
example profile (`small-lan`) as a resource for adding code that adds
a compute node to your copy of the srslte-otalab profile. For
reference, that profile is here:

https://www.powderwireless.net/show/PortalProfiles/small-lan

This profile demonstrates how to allocate nodes and add them to a LAN.
You should modify your profile to add a compute node via a LAN to your
UE nodes using the `small-lan` profile as a guide.

The highly likely reason your profile doesn't execute the startup
scripts is because you changed the paths to these scripts to something
invalid. You have relative paths that will be arbitrarily attached to
whatever directory the startup script is run from. That is almost
certainly not what you want, so you should revert this change and use
absolute paths such as were in the profile script previously.

-Kirk

> --
> You received this message because you are subscribed to the Google Groups "Powder Users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to powder-users...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/powder-users/CAHUWr4X1aM%3DiDNkBnKVJMJT9chfa6dyzMCCYd9%2BiSBDRYzHkQg%40mail.gmail.com.

[Xin Yao]

Dear Kirk,

Thanks for your answer!

I already modified the profile based on the example file you gave me, and it was successfully uploaded to the powder platform last week. Before I changed the path for the initial files to "bin/file_name.sh", I also tried the exact path like "/local/repository/bin/file_name.sh" like the profile script, but the node was also not running the startup scripts. I will try it again, but I wonder if there is any other possible reason that the node cannot run the startup scripts.

Best wishes,

Xin Yao

To view this discussion on the web visit https://groups.google.com/d/msgid/powder-users/CAHhYwBjD8ViyPcAHYzuH7MDyt0sk3af0jKfxX4T%2BZD5nzyJYsQ%40mail.gmail.com.

[Kirk Webb]

The node might not be able to run your script if it doesn't have the
execute bit set, or perhaps the script failed when it ran in some way.
There are logs for these scripts in /var/tmp/startup-N.txt, where "N"
is a number.

I did not see evidence of another compute node being added to your
profile script.

-Kirk

> To view this discussion on the web visit https://groups.google.com/d/msgid/powder-users/CAHUWr4W_d1hX5muSdUv%3D3%3DE00ZvkCtQPG6N99wdVB046VyrFhQ%40mail.gmail.com.