Hey there!
I have a 5-node cluster with a lot of LXC which basically fall into these categories:
Please let me know if this approach would be correct:
Thanks in advance.
I have a 5-node cluster with a lot of LXC which basically fall into these categories:
- Common services (e.g., internal DNS)
- Database servers (e.g. PostgreSQL)
- Application servers (e.g. Python applications running with Gunicorn)
- Web servers (e.g. NGINX as reverse proxy for the application servers).
- Other (e.g. monitoring, analytics).
Please let me know if this approach would be correct:
- Take all LXC in the level 1 and set start/shutdown order to 10. Level 2 would be 20, and up to level 5 with 50.
- All existing containers would have to be updated with a value, otherwise the default order (empty at the moment) would ruin the whole thing.
- Go to the LXC with the highest id of a given level and set the startup delay to 10 seconds (this is not really a handy solutions as they come and go, but I see no other option).
Thanks in advance.