...
- in-paralell - this is the least risk option but requires additional resources
- in-situ - suits single server installations with a single outage window.
Must Check before you start in-situ
Check each of the following items before upgrading in-situ.
- Are you on an older OS? e.g. Centos6, Debian 7 etc.
- upgrade all servers to newer distribution or create new server instances and use the in parallel upgrade
...
- install and migrate to systemd or create new server instances and use the in parallel upgrade
- Does your server _lack_ plenty of extra disk and possible CPU to accommodate both installs
- add extra disk capacity to allow copying of current data
- or use the in parallel upgrade and create new server instances with similar disk capacity
...
- You have a test system and have tested the upgrade process or you have a complete snapshot of your current server with a tested restore
- Complete a test upgrade on a test server before attempting on live system
- Check down time for your service during the upgrade is acceptable, it will take time and if you have a multi server system all servers need to be completed before the complete service is restored.
- An in-paralell build would enable zero downtime and provide an opportunity to test and confirm before migrating.
- Are you new to linux system administration and debugging