fix: Use fixed event dispatcher node ID

[achimnol]

fixes lablup/backend.ai#345

After applying an update including this patch, all users MUST follow the instructions here:

• Shutdown all manager instances.
• Check the result of python -c 'import socket;print(socket.gethostname())' in all manager nodes.
  • If they are all different and kept persistent across reboots, don't need to do anything.
  • If they are not persistent (e.g., includes DHCP-assigned IP addresses) and not unique to each other, set manager.id config value in manager.toml.
  • For development setups on laptops, it is HIGHLY RECOMMENDED to configure manager.id manually because it may change depending on where you are (e.g., different WiFi networks may assign different hostnames).
• Open a redis-cli shell of the Redis server used by agents and managers and execute:
  • select 4 to switch to the event streaming database
  • flushdb to delete all existing (randomized) consumers and pending messages inside their PEL entries
  • quit
  • NOTE: This redis cleanup proceduer is also required when you change the number of manager instances (e.g., the number of manager nodes, the number of manager worker processes) to prevent leaking consumers.
• Restart all manager instances.

[codecov]

Codecov Report

Merging #513 (89cee1f) into main (2f194f9) will increase coverage by 0.00%.
The diff coverage is 100.00%.

❗ Current head 89cee1f differs from pull request most recent head aec1ed8. Consider uploading reports for the commit aec1ed8 to get more accurate results

@@           Coverage Diff           @@
##             main     #513   +/-   ##
=======================================
  Coverage   48.83%   48.84%           
=======================================
  Files          54       54           
  Lines        8969     8970    +1     
=======================================
+ Hits         4380     4381    +1     
  Misses       4589     4589           

Impacted Files	Coverage Δ
src/ai/backend/manager/server.py	59.73% <ø> (ø)
src/ai/backend/manager/config.py	44.03% <100.00%> (+0.17%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 2f194f9...aec1ed8. Read the comment docs.