-
Notifications
You must be signed in to change notification settings - Fork 0
Home
ssh -A root@%vps_ip_address%
For example:
ssh -A root@85.17.45.19
NOTE: all Docker commands must be executed from the ~/tensorflowdocker folder
login into Docker container command line:
docker-compose exec tf bash
show Docker container logs:
docker-compose logs tf
start the jupyter-notebook container:
docker-compose up -d tf
stop the container:
docker-compose stop tf
restart the container:
docker-compose restart tf
Upload file to the virtual machine by scp into ~/tensorflowdocker/data folder:
scp /path/to/file/on/localhost root@%vps_ip_address%:~/tensorflowdocker/data/
Or the same for whole folder:
scp -r /path/to/localhost/folder/ root@%vps_ip_address%:~/tensorflowdocker/data/
~/tensorflowdocker/data/ already mounted into jupyter container, now you could find your data in /opt/data inside the container or from jupyter
This command shall download all data into the current directory on your localhost:
scp -r root@%vps_ip_address%:~/tensorflowdocker/data/ .
Second method: install Midnight Commander file manager (mc) and login into VPS machine through the
F9 menu -> Right -> Shell link… -> root@vps_ip_address%->enter password.
Right pane of mc should list the VPS filesystem; selected files & folders could be copied with F5. Press TAB to switch between panes.

First login into Docker container:
cd ~/tensorflowdocker
docker-compose exec tf bash
Then from within the Docker container standard apt tools are available:
apt-get update && apt-get install %packagename%
apt-cache search %query%
Use NVIDIA SMI command line utility:
# nvidia-smi
Mon Oct 8 10:16:43 2018
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 390.72 Driver Version: 390.72 |
|-------------------------------+----------------------+----------------------+
To open Jupyter built-in terminal, locate an item “Terminal” in the top right “New” drop-down menu:

To connect remote SSH servers from Jupyter Notebook Terminal, install and use ssh/scp tools:
apt install ssh
ssh root@%ip_address%
scp root@%vps_ip_address%:/path/to/file
NOTE: NVIDIA CUDA Toolkit and utilities are available from the Terminal.
For the VPS order follow the link: https://www.reg.ru/cloudvps/order_free?token=hackmoscow2018
NOTE: It is enough of one account per command. After creating the server, the letter with a root-password will be sent to the specified e-mail.
Завершение регистрации в облачных серверах
After registration you will be taken to the servers management panel. Ignore the message «Your account is not protected enough». There is no need for this on Hackathon.
Панель управления серверами
On the «Service's Balance» tab make sure that the bonus balance is equal to 250 bonuses. If there are no bonuses, refer to REG.RU rack.
Вкладка «Баланс услуги»
To create a server, push «Add server» button on «Servers» tab. There will be enough bonuses for three days of work of one Cloud-4 server or two Cloud-3 servers.
Выбор параметров сервера
If you need more resources, then you should reconsider the idea of your project.You can use applications, but make sure that the set of incoming software suits you.
After creating the server, copy the password and save it.
Процесс создания сервера
Click on the server name and go to the server management page.
Страница управления сервером