Only parse a single line from lscpu to get the Cpu(s) count.

[Gullumluvl]

Dear Discourse maintainers,

This is intended to fix an error I was getting with discourse-setup on Debian, from line 261:

bash: 12
scaling*                   2: syntax error in expression (error token is "scaling*                   2")

Caused by the parsing of lscpu command retaining 2 lines instead of just the intended one.

This occurs on my PCs with Debian testing (lscpu from util-linux 2.40.2) and Debian stable (util-linux 2.38.1) because it outputs a line formatted like this (in the "Vendor" section):

CPU(s) scaling MHz:                   20%

(I also checked on a machine with CentOS, there the format is CPU MHz: 2700.000).

To avoid that, I made the regex more specific, but I also added an exit statement in awk. Probably just one of these fixes could be used.

I believe the script would have worked with the version prior to this commit from a month ago: d00c2e7

Best,

[pfaffman]

Maybe this is simpler:

lscpu -p|grep -v "#"|wc -l

Here's one AI recommended:

lscpu | sed -n 's/^CPU(s):\s*//p'

These both work inside of a Debian 12 container and Ubuntu 22.04

[Gullumluvl]

lscpu -p|grep -v "#"|wc -l

This seems best to me, but we must double check with the people using ARM processors, as the commit I mentioned modified this almost exact same line.

Anything expected from me at this point? Sorry I have little experience with contributing to big projects so let me know.

[Gullumluvl]

Just rebased my branch on top of main, and amended the commit with a fix for localization.