Still having problems with Xeoncross' lowendscript

drmike

Greets:

Let me begin by saying that I'm very frustrated by this. Please excuse me.

I'm trying once again to get Xeoncross' lowendscript to work and I'm still having issues. Hoping someone can help.

I've done a bunch of editing to fix the issues that I've had previously. I still can't get my head wrapped around how php is getting invoked though. Looking at the script, I'm seeing lines that both cgi and mvc are being used. In fact it appears to me that the wordpress portion of the script is calling phpcgi while domain is calling mvc. It looks like when I do a top, I'm running phpcgi. I'm going to go with that.

If you think that mvc is better, right now I don't want to hear it. (Excuse me. I'm really ticked. I'm going on like hour 3 on this. I have better things to be doing.)

The include line in the conf file is this:

include /etc/nginx/fastcgi_php;

There is nothing on the server with that file name. Done a updatedb and a locate, gone looking, etc. I'm thinking it sould be:

include fastcgi_params;

Yes? No?

Can someone please kindly help?

Thanks

drmike

I just figured it out. The include should be pointing at:

/etc/nginx/php.conf

It's mentioned further up in the file. Sorry for the trouble.

yomero

lol mike

drmike

Not been a fun day.

And then to boot, our svn/trac provider is kicking our nothing but errors so I can't push out any code.

If that script was on a svn somewhere, I'd submit the fixes. Not setup to do git. And still no clue why the script talks about mvc but doesn't appear to set it up.

yomero

Is just an example I guess (about mvc)

Xeoncross

Hey @drmike, you have my email so feel free to ping me with any questions.

The script sets up two nginx configs for handling passing requests to PHP.

# Sites that have multiple .php file entry points (nginx > 0.7.27) 
/etc/nginx/php.conf

Which is for Wordpress and other systems that have many "entry-points" like wp-admin.php, index.php, otherfile.php, etc.

# MVC frameworks with only a single index.php entry point (nginx > 0.7.27)
/etc/nginx/php.mvc.conf

PHP frameworks use a single index.php file which receive the request and then route it to the correct controller. (CakePHP, CodeIgniter, Zend, etc..)

Depending on what you are running you need to chose one or the other.

Incidentally, I am planing some updates to my script in the near future (testing right now) to move to PHP-FPM and add Postfix/PostgreSQL support. I will also be fixing the domain function.

drmike

I know I have about 8-10 tickets logged about the script.

Part of the problem is that you're referencing mvc but not setting it up in the script.

drmike

And it appears that the setup of itables causes an issue when sftp gets run because I just got locked out of my vps.

I haven't confirmed that so I won't open up a ticket but just wanted to throw that out there.

Xeoncross

Sorry @drmike, I don't get emails when issues are created on lowendadmin's copy of the repository. It's rather silly how github has it setup actually. Here he has to deal with all these issues that don't even apply to his codebase.

sDsB

I'm hijacking this thread to ask a question about the lowendscript.

I'm using it to "strip down" the installation, so I'm only using the "system" and "optimize" commands . However it seems to do the opposite. I start off with 30-ish MB used and end up with 120+ (according to free -m).

Anyone have any idea what's going on, cause ps aux doesn't really show any process using up that much memory.

Xeoncross

After all the changes to the system, there might be extra swap or something that is no longer used - but still reported. You might want to check htop or top and see whats running - and if nothing sticks out then just restart the server to see if that clears everything.

sDsB

Restart does nothing. Server is currently hovering around 112 used.

~# free -m
             total       used       free     shared    buffers     cached
Mem:           128        112         15          0          0        105
-/+ buffers/cache:          7        120
Swap:          128          3        124

Nothing is using up resources, bash is the biggest %mem user with 1.3%.

vmstat agrees with free -m.

Xeoncross

@sDsB, What does ps -aux show?

sDsB

# ps aux
USER       PID %CPU %MEM    VSZ   RSS TTY      STAT START   TIME COMMAND
root         1  0.0  1.0   2608  1440 ?        Ss   23:13   0:00 init
root         2  0.0  0.0      0     0 ?        S    23:13   0:00 [kthreadd/229]
root         3  0.0  0.0      0     0 ?        S    23:13   0:00 [khelper/229]
root         4  0.0  0.0      0     0 ?        S    23:13   0:00 [nfsiod/229]
root       113  0.0  0.7   2180  1000 ?        Ss   23:13   0:00 /usr/sbin/xinetd -dontfork -pidfile /var/run/xinetd.pi
root       116  0.0  0.6   2220   792 ?        Ss   23:13   0:00 cron
root       124  0.0  0.5   1880   680 ?        S    23:13   0:00 /usr/sbin/syslogd
root       169  0.0  0.9   2532  1184 ?        Ss   23:13   0:00 dropbear -i
root       170  0.0  1.3   3096  1736 pts/0    Ss   23:14   0:00 -bash
root       207  0.0  0.7   2476   968 pts/0    R+   23:40   0:00 ps aux

kylix

@sDsB: As you seem to be using OpenVZ you need to look at the first row in column -/+ buffers stating used in free -m.

Xeoncross

@sDsb, http://www.linuxatemyram.com/

sDsB

LOL, thats awesome.

Thanks guys.

yomero

Pfff, another one...

sorry

drmike

How do I bump up the number of allowed connections in iptables? Currently I guess it;s three per this but just a simple sftp connection appears to hit that limit. (Because it's Passive I guess?)

edit: I got it. I edited the config script and then rebooted to get it to load again.

Xeoncross

The iptables script is located at /etc/iptables.up.rules so you can always just edit the file.

iptables -F
vim /etc/iptables.up.rules
iptables-restore < /etc/iptables.up.rules

drmike

Yup, that was the file. Just bumped it up to 5 for the time being.

Surprised that it doesn't affect torrenting. I guess because that's outbound?

Xeoncross

It only limits SSH connections (or anything using port 22).

drmike

Yup, that's the problem since sftp goes through 22.