How to check installation integrity

Checking System requirements

Better option is to install SONM on Ubuntu 16.04 x64 version with autoinstallation script

System requirements:

  • Autoinstall script works only on supported Debian-based OS - better option is to install one.
    • List of supported Linux distributives:
      • Debian 9
      • Ubuntu 16.04 LTS
      • Ubuntu 18.04 LTS
      • Linux Mint 18.1 "Serena"
  • Proprietary GPU drivers must be installed manually - from GPU vendor website(not from repository).
  • If you have other Debian-based OS - install SONM manually from existing packages.

Checking SONM CLI

Warning! Do not leave pass_phrase field blank!

CLI configuration file default location: ~/.sonm/cli.yaml. Keystore vault default location: /etc/sonm/sonm-keystore

Easy way to check CLI installation success - try to open keystore vault: sonmcli login.

List of most common troubles

  • [ERR] Cannot decrypt default key with given pass

    • Solution 1: Check CLI settings: cat ~/.sonm/cli.yaml. Replace incorrect passphrase with correct one via nano nano ~/.sonm/cli.yaml.
    • Solution 2: Change permissions on keystore vault: sudo chmod -R 777 /etc/sonm/sonm-keystore
  • Keystore is empty, generating a new addr(but key exists and located in keystore vault)
    • Solution: Change permissions on keystore vault: sudo chmod -R 777 /etc/sonm/sonm-keystore

Checking SONM Node

Node configuration file default location: /etc/sonm/node-default.yaml. Command to restart node service: : sudo service sonm-node restart.

Easy way to check node installation success - try to get balance on account: sonmcli token balance. If results are look like this, everything is in order:

List of most common troubles

  • [ERR] Cannot read default address from keystore: invalid address 0x specified for parsing

    • Solution 1: sonmcli login, after that select and copy one address from list below, sonmcli login <addr>, enter passphrase
    • Solution 2: sudo chmod -R 777 /etc/sonm/sonm-keystore/state
  • [ERR] Error while dialing dial tcp 127.0.0.1:15030: connect: connection refused
    • Solution: Node is offline. Check if pass_phrase and keystore_path in /etc/sonm/node-default.yaml match the ones in ~/.sonm/cli.yaml. Try to restart node service after applying changes.
  • If none of that worked - get the node log: journalctl -u sonm-node --since "1 hour ago" > node-log.txt and send it to us in Tech Support chat with description of your problem.

    node-log.txt is located in your current working directory.

Checking SONM Worker

Worker configuration file default location: /etc/sonm/worker-default.yaml. Command to restart worker service: sudo service sonm-worker restart.

IMPORTANT! Worker launches several benchmarks on it's first start - GPU redshift benchmark may take about ~1 hour. After restart, worker relaunch net-bandwidth benchmarks. Make sure they are over before sending requests to worker.

Execution of sudo docker ps command will show all containers currently running in your system. Running benchmarks looks like this:

After benchmarking process completion check worker status: sonmcli worker status

List of most common troubles

  • [ERR] cannot get worker status: rpc error: code = Unavailable desc = all SubConns are in TransientFailure, latest connection error: connection error: desc = "transport: Error while dialing dial tcp 127.0.0.1:15030: connect: connection refused"

  • No worker avaiting for confirmation in SONM GUI

    • Solution 1: Make sure if master: value in configuration file: cat /etc/sonm/worker-default.yaml matches the one in SONM GUI. If not - open your config file with sudo nano /etc/sonm/worker-default.yaml, replace master: value with right one and ask for further instructions.
    • Solution 2: Try to restart worker and check out if benchmarks are start to run. If not - get the worker log: journalctl -u sonm-worker --since "1 hour ago" > worker-log.txt and send it to us in Tech Support chat with description of your problem.
  • Failed to create Worker instance: cannot initialize GPU plugin for vendor"radeon": cannot detect required GPU

    • Solution 1: In case you have Nvidia GPU - open your config file with sudo nano /etc/sonm/worker-default.yaml, and replace vendor with the right one in GPU section. This error often occurs because of wrong GPU driver installation. To make sure drivers setuped correctly run nvidia-smi command, command results shoud not be blank. Restart your worker after that.
    • Solution 2: In case you have AMD GPU - install proprietary driver
  • Failed to create Worker instance: cannot initialize GPU plugin for vendor"nvidia": Could not load UVM kernel module. Is nvidia-modprobe installed?
    • Solution: To make sure drivers setuped correctly run nvidia-smi command, command results shoud not be blank. Try to install nvidia-modprobe: sudo apt-get install nvidia-modprobe. Restart your worker after that.

Checking SONM Optimus

Optimus will not create sell plans, if there are no buy orders that fit your hardware, availible for your SONM Worker

Optimus configuration file default location: /etc/sonm/optimus-default.yaml. Command to restart optimus service: sudo service sonm-optimus restart.

Check if Optimus is active systemctl is-active sonm-optimus

List your ask-plans sonmcli worker ask-plan list Wait for optimus to perform optimization - it won't start this process immediately after setup.

If it's active and did not started to create ask-plans - maybe there are no orders for you availiable. Better option is to ask your question here - in Tech Support chat.