LFS201 Lab 3.1 Question #3 seems incorrect

cmullins

In the previous Question 2, we set the limit (soft) to the hard limit (4096).

Question 3 then asks us to set the "HARD" limit to 2048 and verify.
"3. Set the hard limit to 2048 and verify it worked."

My solution would be:

"ulimit -H -n 2048" and of course would fail since you can't set the hard limit lower than the soft limit.

But lab's solution seems to only set the "SOFT" limit:

Lab 3.1 Solution:
"3. $ ulimit -n 2048"
" $ ulimit -n"

Perhaps the question should ask us to set the limit (SOFT) to 2048 as per the solution?

k0dard

Maybe it should be said in the course that -n changes both limits and that specifying -H or -S changes hard or soft limit respectfully...

cmullins

If only the course would mention "If neither -H nor -S is specified, both the soft and hard limits are set." or at least lead us to this conclusion via the lab.

k0dard

Hello cmullins,

Check this out https://forum.linuxfoundation.org/discussion/comment/28989

cmullins

Hi k0dard.

Your post is insightful regarding Question #4.

Unfortunately, my post is regarding Question #3 which explicitly states, "3. Set the hard limit to 2048 and verify it worked" (notice they used the term "hard" limit and yet their solution is setting the soft limit $ ulimit -n 2048".

Again, perhaps the question should ask us to set the limit (SOFT) to 2048 as per the solution?

cmullins

The real crux of the issue here is that the content provided ("Controlling Processes with ulimit") so far in this coursework is lacking vital details such that lowering a limit (soft) also lowers the hard limit even though it explicitly states "Hard Limits -- The maximum value, set only by the root user, that a user can raise the resource limit to."

So when a non root user lowers the soft limit, it also lowers the hard limit but that contradicts their definition of "set only by the root user".

I'm more and more getting the feeling I should ask for a reimbursement of this course and certification because honestly this course is not up to snuff as far as I'm concerned.

coop

This is not strange behaviour in the slightest, and I would maintain the course discussion is not wrong.

All this is similar to what happens when you change a process' priority as a normal user. If you lower the priority (raise the niceness) you can not raise it back to the same level. The change in the hard value is only in the current process (shell) and if you look in another terminal or for a different user you will see the "hard" value does not change.

Many of these commands contain subtleties (and variations due to distribution and sometimes version) that cannot be covered without writing really long sessions that are redundant with man pages, as well as the infinite number of resources on the web; especially on a non-volatile topic like ulimit which does not depend on cutting edge development as it has been around and only slowly evolving for many years and many operating systems.
For example writing industrial strength scripts and code buries the few lines we want to show in a forest of error checks, input data examination etc and diminishes learning.

I don't know what you were expecting as this course is about 700+ pages long and has over 40 sections; going into this level of detail on features you discover through experimentation (as you did rightly here) would bury students, especially since LFS201 is not designed to stand alone as the only thing needed to pass LFCS (at a minimum you need to take LFS101x or equivalent and have plenty of experience). It's not hard to find things you would express differently etc, but that's hardly an indictment.

cmullins

I would prefer knowledge clearly delineated and then tested upon as opposed to brief summaries which lack details and then unspoken subtleties tested upon.

Learning the "hard" way about these subtleties via posting questions to forums seems tedious, I would much rather prefer it taught up front and then tested.

Just because everyone before the mainstream adoption of the internet had to learn this way does not make it ideal anymore.

k0dard

@cmullins said:
Hi k0dard.

Your post is insightful regarding Question #4.

Unfortunately, my post is regarding Question #3 which explicitly states, "3. Set the hard limit to 2048 and verify it worked" (notice they used the term "hard" limit and yet their solution is setting the soft limit $ ulimit -n 2048".

Again, perhaps the question should ask us to set the limit (SOFT) to 2048 as per the solution?

I don't have course access anymore so I don't know which question is which.
My point was that ulimit -n sets BOTH hard and soft limit and you keep insisting that the solution is setting the soft limit...

k0dard

@cmullins said:
So when a non root user lowers the soft limit, it also lowers the hard limit but that contradicts their definition of "set only by the root user".

If you want to lower only the soft limit you need to put -S, I think that's the part you've missed

see man bash and find explanation for ulimit command

Provides control over the resources available to the shell and to processes started by it, on systems that allow such control. The -H and -S options specify that the hard or
soft limit is set for the given resource. A hard limit cannot be increased by a non-root user once it is set; a soft limit may be increased up to the value of the hard limit.
If neither -H nor -S is specified, both the soft and hard limits are set.

luisviveropena

Hi @k0dard ,

Maybe it should be said in the course that -n changes both limits

Yep, "-n" changes both limits. And it's good you found it! Because there are lots of small and tricky things that aren't even in the man pages and "--help". And the only way you will learn these things is by practicing and going further.

Regards,
Luis.