On the systems I administer, I try to setup maintenance cycles. I document a list of things I need to do and perform them during the slowest traffic time I can find. Here is an example list:
- Repair/optimize database tables
- Audit backup operations
- Install any updates
- Restart services and sometimes reboot servers
- Check error logs for anything unusual
- Check file system state
I make sure I have this list open during each maintenance event and modify it as needed.
This has helped ensure stable operations of the systems I administer. I find that I can catch many issues before they affect the client. For example, I see a drive filling up from normal use long before it becomes full. I normally try to schedule this in weekly and/or monthly cycles.