Skip to content

vm.md

Latest commit

 

History

History
202 lines (167 loc) · 7.2 KB

vm.md

File metadata and controls

202 lines (167 loc) · 7.2 KB
Raw

Embeddings.cc Virtual Machine

  • Hostname: embeddings.cs.uni-paderborn.de
  • IP: 131.234.28.19
  • IPv6: [::ffff:83ea:1c13]
  • Created: 2022-02-22
  • OS: Debian GNU/Linux 11 (bullseye) (cat /etc/*-release)
  • Ports open: 443/tcp, 8443/tcp
  • Hardware:
    • CPU: 4x Intel(R) Xeon(R) CPU E5-2695 v3 @ 2.30GHz (cat /proc/cpuinfo)
    • Memory: 32 GB (free -h)
    • Disk: 1007 GB (/dev/sdb, df -h)

Logs

  • Elasticsearch: /data/es8-logs/
  • Elasticsearch garbage collection: /data/elasticsearch-8.3.1/logs/gc.log
  • systemd: sudo journalctl -u embeddings.service -b
  • nginx: cat /var/log/nginx/error.log

Elasticsearch installation

Elasticsearch 8.3.1

  • ES 8.3 installation
    • cd /data
    • wget https://artifacts.elastic.co/downloads/elasticsearch/elasticsearch-8.3.1-linux-x86_64.tar.gz
    • wget https://artifacts.elastic.co/downloads/elasticsearch/elasticsearch-8.3.1-linux-x86_64.tar.gz.sha512
    • shasum -a 512 -c elasticsearch-8.3.1-linux-x86_64.tar.gz.sha512
    • tar -xzf elasticsearch-8.3.1-linux-x86_64.tar.gz
    • rm elasticsearch-8.3.1-linux-x86_64.tar.gz* https://www.elastic.co/guide/en/elasticsearch/reference/8.3/settings.html
  • ES 8.3 settings
    • cp /data/elasticsearch-8.3.1/config/elasticsearch.yml /data/elasticsearch-8.3.1/config/elasticsearch.original.yml
    • nano /data/elasticsearch-8.3.1/config/elasticsearch.yml
    • cluster.name: embcc
    • node.name: embcc-1
    • path.data: /data/es8-data
    • path.logs: /data/es8-logs
    • bootstrap.memory_lock: false
    • network.host: 0.0.0.0
    • http.port: 9200
  • ES 8.3 virtual memory
    • sudo sysctl -w vm.max_map_count=262144 (temp)
    • sudo sysctl vm.max_map_count (check)
    • sudo nano /etc/sysctl.conf (permanently)
      vm.max_map_count=262144
  • ES 8.3 start
    • /data/elasticsearch-8.3.1/bin/elasticsearch -d -p /data/elasticsearch-8.3.1/pid.txt
  • ES 8.3 password
    • /data/elasticsearch-8.3.1/bin/elasticsearch-reset-password --username elastic --interactive
  • Autostart
    • crontab -e
    • @reboot /data/elasticsearch-8.3.1/bin/elasticsearch -d -p /data/elasticsearch-8.3.1/pid.txt
    • sudo shutdown -r 0

Packages / software installation

Webservice installation

  • Initial version, updates in deployment.md
  • Copy code
    • kinit wilke
    • ./scripts/vm-push.sh
    • ssh wilke@embeddings.cs.upb.de
    • . /opt/bashrc.sh
    • sudo mv /tmp/embeddings.cc/ /opt/embeddings.cc/
    • mkdir /opt/embeddings.cc/instance
    • cp /opt/embeddings.cc/config.py /opt/
    • ln -s /opt/config.py /opt/embeddings.cc/instance/config.py
  • Anaconda environment
    • conda create --name embeddings
    • . /opt/anaconda3/etc/profile.d/conda.sh
    • conda activate embeddings
  • Python modules
    • cd /opt/embeddings.cc/
    • conda install -c conda-forge --file requirements.txt
      • ImportError: cannot import name 'soft_unicode' from 'markupsafe'
    • pip install markupsafe==2.0.1
  • Webservice configuration
    • python3 /opt/embeddings.cc/scripts/generate-salt-password.py XXX
    • cp /opt/embeddings.cc/config.py /opt/embeddings.cc/instance/config.py
    • Inserted salt, hash and es-config
  • Start webserver
    • export FLASK_APP=webservice_index
    • export FLASK_RUN_PORT=8008
    • flask run --host=0.0.0.0
    • http://embeddings.cs.uni-paderborn.de:8008/ping --> Status: OK :)
    • New linux user 'embeddings'
      • sudo useradd -m embeddings
      • sudo passwd embeddings
      • sudo -u embeddings -s (also: how to switch to user)
    • sudo ln -s embeddings_cc_e embeddings

Public webservice: Start configuration

uWSGI

[uwsgi]
; This is the uWSGI config, located at /opt/uwsgi.ini
; How to start manually:
;  Log in as user 'embeddings'
;  Change directory to embeddings code root
;  Execute 'uwsgi /opt/uwsgi.ini'

home    = /data/anaconda/
plugins = python3
mount   = /=webservice_public/wsgi.py

socket       = /tmp/embeddingscc.sock
chmod-socket = 666

master             = true
enable-threads     = true
manage-script-name = true
thunder-lock       = true

systemd

  • sudo nano /etc/systemd/system/embeddings.service
  • systemctl enable embeddings.service
  • sudo systemctl start embeddings.service
  • sudo shutdown -r 0
# /etc/systemd/system/embeddings.service

[Unit]
Description=embeddings uwsgi
Requires=network.target
After=network.target

[Service]
WorkingDirectory=/opt/embeddings
User=embeddings
ExecStart=/usr/bin/uwsgi /opt/uwsgi.ini
TimeoutStopSec=30

[Install]
WantedBy=multi-user.target

.bashrc

The folling lines were added for user wilke at conda installation.
Afterwards, they were copied to /home/embeddings/.bashrc.

# >>> conda initialize >>>
# !! Contents within this block are managed by 'conda init' !!
__conda_setup="$('/data/anaconda/bin/conda' 'shell.bash' 'hook' 2> /dev/null)"
if [ $? -eq 0 ]; then
    eval "$__conda_setup"
else
    if [ -f "/data/anaconda/etc/profile.d/conda.sh" ]; then
        . "/data/anaconda/etc/profile.d/conda.sh"
    else
        export PATH="/data/anaconda/bin:$PATH"
    fi
fi
unset __conda_setup
# <<< conda initialize <<<

Misc

  • Initialize bash (prompt and anaconda): . /opt/bashrc.sh
  • Kerberos Single-Sign-On: kinit <imt-username>
  • Prompt (see wiki.ubuntuusers.de):
    PS1='\[\e]0;\u@\h: \w\a\]${debian_chroot:+($debian_chroot)}\[\033[01;32m\]\u@\h\[\033[00m\]:\[\033[01;34m\]\w\[\033[00m\]\$ '