nvidia/nls
2
0
Fork 0
Code Issues Pull Requests Projects Activity

updated notes

Browse Source
This commit is contained in:
Oscar Krause 2025-03-19 20:53:49 +01:00
parent b8b6ae1c87
commit 6ea36bd340
4 changed files with 448 additions and 77 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

269
DEBUG.md Normal file
View File

File diff suppressed because one or more lines are too long

37
Database.md Normal file
View File

@ -0,0 +1,37 @@
# Database structure
[__TOC__]
# `si_<id>`
Here is the most data stored.
# `request_routing'
Just basic uninteresting data.
## `request_routing.service_instance`
| xid | org_name |
|----------------------------------------|--------------------------|
| `10000000-0000-0000-0000-000000000000` | `lic-000000000000000000` |
- `xid` is used as `SERVICE_INSTANCE_XID`
## `request_routing.license_allotment_service_instance`
| xid | service_instance_xid | license_allotment_xid |
|----------------------------------------|----------------------------------------|----------------------------------------|
| `90000000-0000-0000-0000-000000000001` | `10000000-0000-0000-0000-000000000000` | `80000000-0000-0000-0000-000000000001` |
- `xid` is only a primary-key and never used as foreign-key or reference
- `license_allotment_xid` must be used to fetch `xid`'s from `request_routing.license_allotment_reference`
## `request_routing.license_allotment_reference`
| xid | license_allotment_xid |
|----------------------------------------|----------------------------------------|
| `20000000-0000-0000-0000-000000000001` | `80000000-0000-0000-0000-000000000001` |
- `xid` is used as `scope_ref_list` on token request

108
README.md
View File

@ -1,86 +1,40 @@
# NLS # NLS - NVIDIA License System
[[_TOC_]] - https://docs.nvidia.com/license-system/latest/index.html
# Docker [__TOC__]
## Load Images # Installation
First copy `nls-X.Y.Z-bios.zip` to `/opt/nls-X.Y.Z-bios.zip` with SFTP/SSH. ## Docker
```shell 1. Install Docker on your system where NLS should run
# set version 2. Download latest release (*date of writing `v3.4.1`*)
export VERSION=3.4.1 3. Copy archive (`nls-X.Y.Z-bios.zip`) to your Docker-Server
# extract archive 4. Set version \
unzip nls-$VERSION-bios.zip `export VERSION=3.4.1`
# goto directory 5. Extract archive \
cd nls-$VERSION-bios `unzip nls-$VERSION-bios.zip`
``` `cd nls-$VERSION-bios`
6. Import Images into Docker \
`docker load --input dls_appliance_$VERSION.tar.gz` \
`docker load --input dls_pgsql_$VERSION.tar.gz` \
`docker load --input dls_upgrade_$VERSION.tar.gz` (optional)
7. Goto `docker` directory
`cd docker`
8. Create Envirnoment-File
`touch .env`
9. Add the following line to this file
`echo "DLS_PUBLIC_IP=<your-ipv4>" > .env`
10. Start Docker-Stack
`docker compose up -d`
11. Show running containers \
`docker container ls`
Import Images into Docker: # Reverse Engineering Notes
```shell Notes about accessing file structure and database can be found [here](Reverse%20Engineering%20Notes.md).
# appliance
docker load --input dls_appliance_$VERSION.tar.gz
# postgres
docker load --input dls_pgsql_$VERSION.tar.gz
# upgrade-image (optional)
docker load --input dls_upgrade_$VERSION.tar.gz
```
## Start service # Debug Information
```shell Debug information about licensing process, token-exchange and certificates can be found [here](DEBUG.md).
cd docker
docker compose up -d
````
## Dump Filestructure
Prepare output:
```shell
mkdir /tmp/out
```
Show running containers:
```shell
docker container ls
```
Copy directory structure (using container-id from `dls:appliance_$VERSION`)
```shell
docker cp -a <container-id>:/venv/lib /tmp/out
# Successfully copied ...MB to /tmp/out/
```
Now you have the required directory structure in `/tmp/out/lib/python?.??/site-packages`.
## Re-Tag Images and upload to this registry
Show Images:
```shell
docker image ls
````
Tag Images:
```shell
docker image tag <image-id-dls_appliance> registry.git.collinwebdesigns.de/nvidia/nls/appliance:$VERSION
docker image tag <image-id-dls_pgsql> registry.git.collinwebdesigns.de/nvidia/nls/pgsql:$VERSION
docker image tag <image-id-dls_upgrade> registry.git.collinwebdesigns.de/nvidia/nls/upgrade:$VERSION
```
Upload Images:
```shell
# Login to private registry
docker login registry.git.collinwebdesigns.de
# Push images
docker image push registry.git.collinwebdesigns.de/nvidia/nls/appliance:$VERSION
docker image push registry.git.collinwebdesigns.de/nvidia/nls/pgsql:$VERSION
docker image push registry.git.collinwebdesigns.de/nvidia/nls/upgrade:$VERSION
```

111
Reverse Engineering Notes.md Normal file
View File

@ -0,0 +1,111 @@
# Reverse Engineering Notes
For this research, NLS Docker-Setup is used.
- More about Docker Images https://git.collinwebdesigns.de/nvidia/nls
[__TOC__]
# Appliance
## Get / Copy file structure
1. Log in to your Docker-Server
2. Create a target directory (e.g. `mkdir /opt/nls-files`)
3. Get the Container-ID of the NLS-Appliance (`docker ps`)
4. Copy files from container \
`docker cp -r <container-id->`
## About configuration data
- Most variables and configs are stored in `/var/lib/docker/volumes/configurations/_data`.
- Config-Variables are in `etc/dls/config/service_env.conf`.
## NLS Logs
Logs are found in `/var/lib/docker/volumes/logs/_data`.
Most interesting logs are:
- `fileInstallation.log`
- `serviceInstance.log`
## File manipulation
- Files can be copied with `docker cp <container-id>:/venv/... /opt/localfile/...`.
- Files can be directly edited via Docker-Volume mounts
- see `df -h` (one is nls, the other postgres container)
```
overlay 16G 11G 5.6G 66% /var/lib/docker/overlay2/<hash>/merged
overlay 16G 11G 5.6G 66% /var/lib/docker/overlay2/<hash>/merged
```
- then you can edit files with e.g. `nano venv/lib/python3.12/site-packages/...`
**After you edit any file directly, service needs to be restarted** `docker restart <container-id>`.
# Database
## DB Access
Valid users are `dls_writer` and `postgres`.
1. Log in to your Docker-Server
2. Get the Container-ID of the NLS-Postgres (`docker ps`)
3. Use psql \
`docker exec -it <container-id> psql -h localhost -U postgres`
4. (optional) Create a superuser for external access \
`CREATE USER admin WITH LOGIN SUPERUSER PASSWORD 'admin';`
5. Add exposed port to `docker-compose.yaml`
5.1 Add `ports: [ '5432:5432' ]` into `...`-section
## Table Structure
Some information about Database / Table structure can be found [here](Database.md).
# Logging / Stack Trace
- https://docs.nvidia.com/license-system/latest/nvidia-license-system-user-guide/index.html#troubleshooting-dls-instance
# Nginx
- NGINX uses `/opt/certs/cert.pem` and `/opt/certs/key.pem`
# Other tools / files
Other tools / files which may can helpful, but not known for what they are used.
**Appliance**
- `/etc/dls/config/site_key_uri.bin`
- `/etc/dls/config/dls_db_password.bin`
**Database**
- `/etc/dls/config/decryptor/decryptor`
# Usefully commands on Client
## Check licensing status
- `nvidia-smi -q | grep "License"`
**Output**
```
vGPU Software Licensed Product
License Status : Licensed (Expiry: 2023-1-14 12:59:52 GMT)
```
## Track licensing progress
- NVIDIA Grid Log: `journalctl -u nvidia-gridd -f`
```
systemd: Started NVIDIA Grid Daemon.
nvidia-gridd: Configuration parameter ( ServerAddress ) not set
nvidia-gridd: vGPU Software package (0)
nvidia-gridd: Ignore service provider and node-locked licensing
nvidia-gridd: NLS initialized
nvidia-gridd: Acquiring license. (Info: license.nvidia.space; NVIDIA RTX Virtual Workstation)
nvidia-gridd: License acquired successfully. (Info: license.nvidia.space, NVIDIA RTX Virtual Workstation; Expiry: 2023-1-29 22:3:0 GMT)
```