@marcinozga said in Linux partitioning:
@travisdh1 said in Linux partitioning:
@marcinozga said in Linux partitioning:
@travisdh1 said in Linux partitioning:
@marcinozga said in Linux partitioning:
@travisdh1 said in Linux partitioning:
@marcinozga said in Linux partitioning:
@JaredBusch said in Linux partitioning:
@marcinozga said in Linux partitioning:
Last option on that screen. Why? Because if your root partition fills up, your system won't boot. Having /var on separate partition reduces that risk. Think of a bad app creating gigabytes of log entries, or writing junk to database.
Better solution, don't use bad apps
You never know if an app or kernel update doesn't go nuts. Here's an example of bad kernel module causing writing gigs of logs in minutes: https://unix.stackexchange.com/questions/195360/my-var-log-is-mysteriously-filling-up-gbs-in-minutes-any-cure-before-i-re-ins
Sure it's possible, but how many times do we see this sort of thing in normal use day to day? This is the first one I've heard of in years.
Used to happen more often because bad programming was more common.
Take Ubuntu for example, it will fill up /boot after a few kernel upgrades, so anything is possible.
That's a known badly implemented system. IE bad programming. Don't use it.
Also, it has nothing to do with /var filling up.
It's just an example of partition filling up, doesn't matter which one. It proves the point that it does happen.
In your example, they are causing a new problem because they've implemented your solution to the problem.
No, it would've had the same outcome if /boot wasn't on separate partition, but under root. Eventually it would've filled up entire disk with old kernel versions. The main point is to prevent root partition from filling up, not any other mount point.
That's not really true. That's making a lot of assumptions. In reality...
It would not likely ever fill up. The scales are so different. This is where creating a separate /boot actually CREATES the risk. The very pattern you are trying to follow is the one causing the issue.
It has lots and lots of chances for humans AND monitoring systems to catch and resolve it long before becoming an issue. This should have years, typically, to be flagged rather than filling up in a day or two.