Not so random musings
There are some things that annoy me. One of those is the answer of "Because that's how we do it" when asked why something is done that way.
Lately one of my gripes has been linux partitioning. When I first started using linux, you would partition a disk with /, /boot, [swap] and probably a /usr/local. The reason was simple, the boot partition had to be under 2G to be bootable on most systems. Swap would be double the RAM and usually we would assign /usr/local to the main user of the system and they'd use that as their main storage space.
In later years, we didn't bother with /boot. [swap] got increased to 8G; / was made to be about 50G which left most of the disk as a data partition which could be used by the user.
Sure, if we had specific needs we would set up specific partitioning, but by and large /, [swap] and some variation on /mnt/data covered it.
Then I started here. Every machine is setup the same. Ok, good, standard partitioning.
/
/boot
/usr
/usr/local
/usr/bin
/tmp
/var
/var/log
[swap]
And the partitions are tiny - I think /usr/local is 2G, everything else is 1G down to a few hundred meg. I asked why. The answer? "That's the way we do it."
Actually, no, that's the way Solaris does it. Why does Solaris do it that way? Because originally they only shipped with tiny disks and they needed to spread the file system across multiple disks to make it work. The first Sparcs I worked on had 200Mb disks, usually four or five of them.
Sure, if you're doing something special, spread the data around, but the vast majority of computers now only have a single disk. Putting / on one partition and /tmp on another isn't going to make a difference, it is still the same head that has to read and write the data.
I've also been told it is so the system or a user can't fill up a disk and stop the system. Umm, yes they can. Sure, the won't (or shouldn't be able to) fill /boot or /usr, but they'll just fill up /tmp. I've had users DOS themselves because they've generated 20Gb error dumps to /var/log. That is one reason we made those partitions big - disks are cheap and in most cases you've got 1Tb+ just sitting there. You're using 20Gb (closer to 8Gb in this case) for the system, why waste the other 980Gb? Let the user access it.
If you're paranoid about them going hog wild, put on quotas, that's what they're for.
I'm sort of bemused that a lot of the problems I've seen lately is because the partitions are so small. Sure, they're LVM partitions, so you can resize them, but why put those restrictions on in the first place?
And don't get me started on networks... or scripts. I spent the last hour at work trying to figure out a script. The scripts are very clever, but that is also their downfall. Script A calls script B which draws in a function from script D, but it might use script C on even Sundays...
K.I.S.S. Keep it simple...
Anyway, just my current rant. I need to breathe deeper and think calming thoughts.
Lately one of my gripes has been linux partitioning. When I first started using linux, you would partition a disk with /, /boot, [swap] and probably a /usr/local. The reason was simple, the boot partition had to be under 2G to be bootable on most systems. Swap would be double the RAM and usually we would assign /usr/local to the main user of the system and they'd use that as their main storage space.
In later years, we didn't bother with /boot. [swap] got increased to 8G; / was made to be about 50G which left most of the disk as a data partition which could be used by the user.
Sure, if we had specific needs we would set up specific partitioning, but by and large /, [swap] and some variation on /mnt/data covered it.
Then I started here. Every machine is setup the same. Ok, good, standard partitioning.
/
/boot
/usr
/usr/local
/usr/bin
/tmp
/var
/var/log
[swap]
And the partitions are tiny - I think /usr/local is 2G, everything else is 1G down to a few hundred meg. I asked why. The answer? "That's the way we do it."
Actually, no, that's the way Solaris does it. Why does Solaris do it that way? Because originally they only shipped with tiny disks and they needed to spread the file system across multiple disks to make it work. The first Sparcs I worked on had 200Mb disks, usually four or five of them.
Sure, if you're doing something special, spread the data around, but the vast majority of computers now only have a single disk. Putting / on one partition and /tmp on another isn't going to make a difference, it is still the same head that has to read and write the data.
I've also been told it is so the system or a user can't fill up a disk and stop the system. Umm, yes they can. Sure, the won't (or shouldn't be able to) fill /boot or /usr, but they'll just fill up /tmp. I've had users DOS themselves because they've generated 20Gb error dumps to /var/log. That is one reason we made those partitions big - disks are cheap and in most cases you've got 1Tb+ just sitting there. You're using 20Gb (closer to 8Gb in this case) for the system, why waste the other 980Gb? Let the user access it.
If you're paranoid about them going hog wild, put on quotas, that's what they're for.
I'm sort of bemused that a lot of the problems I've seen lately is because the partitions are so small. Sure, they're LVM partitions, so you can resize them, but why put those restrictions on in the first place?
And don't get me started on networks... or scripts. I spent the last hour at work trying to figure out a script. The scripts are very clever, but that is also their downfall. Script A calls script B which draws in a function from script D, but it might use script C on even Sundays...
K.I.S.S. Keep it simple...
Anyway, just my current rant. I need to breathe deeper and think calming thoughts.