Yes, that's a lot. I don't know what of that is default on CentOS and what is autostarted or required by Webmin.

But if you don't need (sending/receiving) mail, then you should deactivate postfix and especially clamav which is a resource hog (well, at least if used through amavis). If I'm not mistaken you also don't need saslauth and I'm guessing mailman. Although, I don't know if CentOS requires active postfix for local notifications delivery? Also, I don't know Linode's CentOS image, but afaik "vanilla" CentOS installation comes with a lot of unnecessary thingies like bluetooth etc… Oh, and you don't really need Bind (named).

Unless, as I said, it is all required by Webmin, and iirc it is.

Just for comparison, my Gentoo node hosting a few low traffic sites (up to a hundred thousand hits per month and a few thousand email transactions, in total) on lighttpd, php (3 processes per site), postgresql, postfix, dovecot, spamassassin, vsftpd, ssh, memcached, rarely exceeds 100MB (and then only due spamd bloat). Upon boot, memory usage is around 70MB.

Because Apache is a hog an IMHO should be avoided unless you require modules not available under other webservers, or per directory .htaccess (in short, if you're not providing a shared hosting service), and Webmin may be cute clicky-clicky config-fest but in two years of my experience in running a VPS, I never really needed it. It is maybe useful for administering a number of domains, if cPanel is too expensive.

I prefer gentoo, install only what I need, configure it per my liking and call it a day. I have my own set of tools and scripts for easy administration and ready-made config files for services that I use.

But that's me. YMMV. :wink:

Apache is pwning your memory usage as said above. Limit your httpd server load via httpd.conf depending on how much you think is necessary.

Thanks for the informative and timely advice,

That comparison is very interesting, it's good to hear real world numbers on the resources that a well tuned site will require. Alas, I'm new to VPS hosting and my site is not overly busy (around 3000 hits to the blog and 6000 hits to the forum per day).

I went with webmin as the clicky admin interface is great for my sanity as I learn linux, and I'm hoping performance will be sustainable for the short to medium term with the current setup. By the time I get my head around the full barebones power of a VPS and then switch over to an optimised instance.

Given these traffic estimates (actually they're our busiest days on record), if I disable/uninstall clamav, postfix, Bind, saslauth, and mailman, then tell apache behave would this be a viable platform for my needs?

I don't know if I really require Apache, though I do need mod_rewrite to run the permalinks in wordpress. Should I be looking to a lighter web server from the start, or perhaps make that the first thing to optimise when performance suffers.

Finally, are there any quick pointers on how to remove the bloatware?

Cheers,

Moses.

I just found Webmin's startup options page, seems I can disable many of these from this screen

~~http://www.greenandgoldrugby.com/upload/startup.JPG" />

For now I've disabled mailman, named (BIND) and saslauthd. Any others I can get rid of?~~

There's a decent chance that tool is "pulling your leg" with regards memory usage… the best way to check it is to ssh in and use the "free" command:

$ free -m
             total       used       free     shared    buffers     cached
Mem:           348        333         15          0          3        219
-/+ buffers/cache:        109        238
Swap:          127         26        101

On the "Mem:" line, it says I'm using 333MB of RAM… however, if you look at the line after that (which subtracts cached stuff), I'm only using 109MB! For some odd reason, most tools that display memory usage like to display the heart-attack total. I believe this is a vast conspiracy by memory manufacturers, but I digress.

This page has a good explanation of the whole thing: http://www.linuxatemyram.com/

Just disabled the above startup processes and applied these two changes
> Two tweaks to reduce memory consumption:

1) If not needed, disable innodb and berkley database by adding the following code to /etc/my.cnf in the [mysqld] section and reboot mysql

skip-innodb

skip-bdb

2) Reduce Apache's memory usage by adding the following to httpd.conf

MaxSpareServers 1

MinSpareServers 1

StartServers 1

Webmin now reports that 86.08MB RAM is now being used ;)

Thanks hoopycat, the free command reports

> > free -m

total used free shared buffers cached

Mem: 360 177 182 0 7 81

-/+ buffers/cache: 89 271

Swap: 511 0 511

@Moses:

For now I've disabled mailman, named (BIND) and saslauthd. Any others I can get rid of?

I'm running CentOS 5.3, and have these services enabled:

$ chkconfig --list | grep 3:on
atd                0:off   1:off   2:off   3:on    4:on    5:on    6:off
auditd             0:off   1:off   2:on    3:on    4:on    5:on    6:off
crond              0:off   1:off   2:on    3:on    4:on    5:on    6:off
haldaemon          0:off   1:off   2:off   3:on    4:on    5:on    6:off
httpd              0:off   1:off   2:on    3:on    4:on    5:on    6:off
ip6tables          0:off   1:off   2:on    3:on    4:on    5:on    6:off
iptables           0:off   1:off   2:on    3:on    4:on    5:on    6:off
irqbalance         0:off   1:off   2:on    3:on    4:on    5:on    6:off
kudzu              0:off   1:off   2:off   3:on    4:on    5:on    6:off
mcstrans           0:off   1:off   2:on    3:on    4:on    5:on    6:off
mdmonitor          0:off   1:off   2:on    3:on    4:on    5:on    6:off
messagebus         0:off   1:off   2:off   3:on    4:on    5:on    6:off
microcode_ctl      0:off   1:off   2:on    3:on    4:on    5:on    6:off
mysqld             0:off   1:off   2:on    3:on    4:on    5:on    6:off
named              0:off   1:off   2:on    3:on    4:on    5:on    6:off
network            0:off   1:off   2:on    3:on    4:on    5:on    6:off
readahead_early    0:off   1:off   2:on    3:on    4:on    5:on    6:off
restorecond        0:off   1:off   2:on    3:on    4:on    5:on    6:off
sendmail           0:off   1:off   2:on    3:on    4:on    5:on    6:off
sshd               0:off   1:off   2:on    3:on    4:on    5:on    6:off
syslog             0:off   1:off   2:on    3:on    4:on    5:on    6:off
vsftpd             0:off   1:off   2:on    3:on    4:on    5:on    6:off
xinetd             0:off   1:off   2:on    3:on    4:on    5:on    6:off

And it's only taking about 100MB memory:

$ free
             total       used       free     shared    buffers     cached
Mem:        553180     491648      61532          0      58404     332188
-/+ buffers/cache:     101056     452124
Swap:       393208         32     393176

And, I could probably turn off several of those, as they aren't doing anything useful.

Basically, what you can do, is search to find out what each of those services do that are running on your VPS. If you don't need them, then turn them off.

I'm guessing, based on what you've said, that you may not need these: clamd-wrapper, named, netfs, postfix, proftpd.

And, as far as I know, these aren't useful on the VPS: lm_sensors, lvm2-monitor, rawdevices

Thanks Kipbond,

I've removed those from the startup list with the exception of proftpd which I believe I use for transferring files to/from the VPS. When the system is running and stable I guess I could turn it off and only switch it on when I have files to transfer…

Strangely enough memory usage has now gone up to 125MB!

Will reboot and see how it comes back up…

You could/should use sftp to transfer files (which will use ssh). It's more secure, and then it's one less service to run on the server.

Cheers, I am using sftp.. didn't realise it was a different process though…. proftpd is off ;)

You can use "top" to get a better view of what is using your memory. Apache may start with only 1-2 processes (if you configured it that way), but then as you get more connections, it starts new processes. I'm not sure if Wordpress works with them, but you might check into using Lighttpd or Nginx if Apache ends up proving to be too much of a memory hog for you.

Thanks again KipBond,

The system is running on a much smaller footprint now, around 120MB seems the norm.

I'll get some users on it tonight for some load testing. Anything I should monitor during the load testing? Linode monitors CPU/Network/IO for me, I guess I shoudl run a few top's through the testing window to see where the RAM is being used….. anything else to have a play with?

I'm thinking I should set Apache to be more in tune with the requirements, currently I have set

> MaxSpareServers 3

MinSpareServers 1

StartServers 1

ServerLimit 50

MaxClients 50

MaxRequestsPerChild 2000

StartServers 2

MaxClients 150

MinSpareThreads 25

MaxSpareThreads 75

ThreadsPerChild 25

MaxRequestsPerChild 0

and I just turned off most of the modules as follows:
> #LoadModule authbasicmodule modules/modauthbasic.so

LoadModule authdigestmodule modules/modauthdigest.so

LoadModule authnfilemodule modules/modauthnfile.so

LoadModule authnaliasmodule modules/modauthnalias.so

LoadModule authnanonmodule modules/modauthnanon.so

LoadModule authndbmmodule modules/modauthndbm.so

LoadModule authndefaultmodule modules/modauthndefault.so

LoadModule authzhostmodule modules/modauthzhost.so

LoadModule authzusermodule modules/modauthzuser.so

LoadModule authzownermodule modules/modauthzowner.so

LoadModule authzgroupfilemodule modules/modauthzgroupfile.so

LoadModule authzdbmmodule modules/modauthzdbm.so

LoadModule authzdefaultmodule modules/modauthzdefault.so

LoadModule ldapmodule modules/modldap.so

LoadModule authnzldapmodule modules/modauthnzldap.so

LoadModule includemodule modules/modinclude.so

LoadModule logconfigmodule modules/modlogconfig.so

LoadModule logiomodule modules/modlogio.so

LoadModule envmodule modules/modenv.so

LoadModule extfiltermodule modules/modextfilter.so

LoadModule mimemagicmodule modules/modmimemagic.so

LoadModule expiresmodule modules/modexpires.so

LoadModule deflatemodule modules/moddeflate.so

LoadModule headersmodule modules/modheaders.so

LoadModule usertrackmodule modules/modusertrack.so

LoadModule setenvifmodule modules/modsetenvif.so

LoadModule mimemodule modules/modmime.so

LoadModule davmodule modules/moddav.so

LoadModule statusmodule modules/modstatus.so

LoadModule autoindexmodule modules/modautoindex.so

LoadModule infomodule modules/modinfo.so

LoadModule davfsmodule modules/moddavfs.so

LoadModule vhostaliasmodule modules/modvhostalias.so

LoadModule negotiationmodule modules/modnegotiation.so

LoadModule dirmodule modules/moddir.so

LoadModule actionsmodule modules/modactions.so

LoadModule spelingmodule modules/modspeling.so

LoadModule userdirmodule modules/moduserdir.so

LoadModule aliasmodule modules/modalias.so

LoadModule rewritemodule modules/modrewrite.so

LoadModule proxymodule modules/modproxy.so

LoadModule proxybalancermodule modules/modproxybalancer.so

LoadModule proxyftpmodule modules/modproxyftp.so

LoadModule proxyhttpmodule modules/modproxyhttp.so

LoadModule proxyconnectmodule modules/modproxyconnect.so

LoadModule cachemodule modules/modcache.so

LoadModule suexecmodule modules/modsuexec.so

LoadModule diskcachemodule modules/moddiskcache.so

LoadModule filecachemodule modules/modfilecache.so

LoadModule memcachemodule modules/modmemcache.so

LoadModule cgimodule modules/modcgi.so

LoadModule versionmodule modules/modversion.so

Webmin can be configured to work with the services you want, the way you want. It doesn't really "need" much of anything, doesn't install anything and doesn't autostart anything unless you configure it to do so. That's it's beauty compared to the "my way or no way" panels like cPanel.

I have a Linode 540 with CentOS 5, Webmin+Virtual, Apache, MySQL, Postfix, Dovecot, etc. It hosts around 30 domains and maybe 60 email users. Memory usage is 100-125 MB.

I'd suggest you…

• Disable the services you don't need. If you're not sure what they do, read up on them or ask. e.g., Mailman definitely can go, and Bind also if you're going to use Linode's DNS manager.

• Look into MySQL memory optimization. There is likely a big saving you can realize there.

• By all means restrict Apache, but don't strangle the poor thing. Your sites will be happier with more than one Apache process running. Try setting those *Servers settings in the 3-6 range and set an upper limit (MaxClients) to maybe 20. Then monitor and see how things go.

Apache prefork should be able to handle 10000 hits/day with no trouble. Here are the settings on a CentOS Linode 360 I have running MediaWiki:

 <ifmodule prefork.c="">StartServers       8
MinSpareServers    2
MaxSpareServers   10
ServerLimit      256
MaxClients        25
MaxRequestsPerChild  400</ifmodule> 

Memory usage:

             total       used       free     shared    buffers     cached
Mem:           360        309         50          0         28         88
-/+ buffers/cache:        193        167
Swap:          255          0        255

It's been up for 43 days and hasn't hit swap yet.

From your list of enabled services, I have haldaemon, ip6tables, kudzu, mcstrans, and messagebus turned off with no apparent ill effect. Note that I'm not using IPv6, SELinux, or Webmin, so your mileage may vary.

Thanks Lance, I'll give your Apache settings a go and see how it runs tonight.

Thanks all for your help so far, it's definately running on a much smaller footprint now so we'll see how it handles some load…

@Moses:

I guess I shoudl run a few top's through the testing window to see where the RAM is being used..

Yes. And hit "M" (capital) when top is running to sort by RAM usage.

I've made big progress this weekend. After much deliberation I decided to stick it out and get nginx firing. During this process I found some excellent guides, however most were for Ubuntu so I ended up trashing the CentOS build and am now running Hardy.

With a massive shout out to The_Guv for his excellent guides, I'm running a clean and mean instance of Ubuntu, nginx, PHP5/fastcgi/xcache and the site is flying… well as quick as I could expect from Sydney anyway.

Now I can use that RAM saved by nginx for caching PHP and MYSQL… as soon as I work out how!

I've got the blog and forums firing now, and will probably migrate the live DNS tonight or tomorrow. Here's the top stats sorted by MEM of the system sitting idle:

> top - 18:06:38 up 7 min, 1 user, load average: 0.04, 0.03, 0.00Tasks: 83 total, 1 running, 82 sleeping, 0 stopped, 0 zombie

Cpu(s): 0.0%us, 0.0%sy, 0.0%ni,100.0%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st

Mem: 368864k total, 185160k used, 183704k free, 4408k buffers

Swap: 393208k total, 0k used, 393208k free, 97204k cached

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND

2668 www-data 15 0 61792 31m 13m S 0 8.8 0:00.72 php-cgi

2669 www-data 16 0 55440 22m 11m S 0 6.3 0:01.07 php-cgi

2287 mysql 15 0 124m 17m 5004 S 0 4.8 0:00.11 mysqld

2667 www-data 15 0 48080 14m 9696 S 0 4.0 0:00.22 php-cgi

2670 www-data 18 0 59008 12m 6516 S 0 3.5 0:00.04 php-cgi

2617 www-data 21 0 46560 9168 6120 S 0 2.5 0:00.06 php-cgi

2671 www-data 18 0 46824 4708 1300 S 0 1.3 0:00.00 php-cgi

2672 moses 17 0 5572 3068 1404 S 0 0.8 0:00.12 bash

2609 www-data 15 0 4916 1636 752 S 0 0.4 0:00.00 nginx

2107 root 16 0 2476 1168 904 S 0 0.3 0:00.00 login

2697 moses 15 0 2220 1076 832 R 0 0.3 0:00.02 top

2610 www-data 15 0 4596 1064 436 S 0 0.3 0:00.00 nginx

2611 www-data 15 0 4596 1064 436 S 0 0.3 0:00.00 nginx

2612 www-data 15 0 4596 1064 436 S 0 0.3 0:00.00 nginx

2166 klog 17 0 2128 1024 392 S 0 0.3 0:00.00 klogd

2187 root 18 0 5220 968 616 S 0 0.3 0:00.00 sshd

1 root 15 0 1956 964 528 S 0 0.3 0:00.00 init

Do the above figures look reasonable? Are there any decent load testing apps that I could use to stress the site and see how she rolls?

Thanks for all the advice to date, it's been educational!

You can run ab (apachebench) to load test the site:

http://www.claytonlz.com/index.php/2008 … -to-nginx/">http://www.claytonlz.com/index.php/2008/10/why-i-switched-to-nginx/

http://httpd.apache.org/docs/2.0/programs/ab.html

Thanks Waldo, those ab tests were great, managed to serve 99% of 10,000 pages < 5 secs!

I did my big server move tonight and it all seems to have gone smoothly so far, and damn it FLIES! Well, it could well slow down a lot once the users get on there, but for now I'm loving it.

I've noticed a bunch of permissions related errors, and have been slowly rectifying these with chmod's. Is there a way to export all the chmod's from my old (dreamhost) server to the VPS? Well that's assuming they were correct in the first place, but it's a starting point at least.

Also, I thought I'd update Wordpress however the automated login seems to not be working. I've set it to use IP.ADD.RE.SS:PORT as well as my SFTP details (that work in FileZilla from home) and get:
> Error: There was an error connecting to the server, Please verify the settings are correct.

Finally, I've got huge amounts of RAM free, should I be looking to increase mysql cache or something to use some of this up?
> $ free -m

total used free shared buffers cached

Mem: 360 256 103 0 30 109

-/+ buffers/cache: 116 243

Swap: 383 0 383

$ top

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND

2429 www-data 18 0 63080 39m 20m S 0 11.1 0:12.34 php-cgi

2430 www-data 16 0 55416 28m 16m S 0 7.8 0:14.58 php-cgi

2432 www-data 16 0 52356 24m 15m S 0 6.8 0:12.81 php-cgi

2433 www-data 16 0 51552 24m 15m S 0 6.7 0:13.61 php-cgi

2431 www-data 16 0 51548 22m 14m S 0 6.3 0:10.99 php-cgi

2271 mysql 15 0 158m 20m 5200 S 0 5.8 0:03.27 mysqld

2385 www-data 22 0 46560 9188 6136 S 0 2.5 0:00.03 php-cgi

2562 root 15 0 6816 4952 1544 S 0 1.3 0:00.01 munin-node

2685 moses 15 0 4440 3024 1268 S 0 0.8 0:00.11 bash

2563 root 16 0 7808 2376 1940 S 0 0.6 0:00.03 sshd

4092 root 16 0 7804 2224 1800 S 0 0.6 0:00.04 sshd

11881 postfix 16 0 5440 1936 1584 S 0 0.5 0:00.00 smtp

Cheers,

Moses.

@Moses:

Finally, I've got huge amounts of RAM free, should I be looking to increase mysql cache or something to use some of this up?

SMF has memory caching capabilities, with memcached, APC and I believe even XCache, the last two also being opcode cachers. I'm assuming you're using one of them.

I'm getting an nginx error when a user posts a comment to the blog, though the comment does still post…

> 502 Bad Gateway

nginx/0.7.62

Should I have chown'ed anything to www-data

Here's my nginx.conf
> user www-data www-data;

worker_processes 4;

events {

worker_connections 1024;

}

http {

include mime.types;

default_type application/octet-stream;

sendfile on;

tcp_nopush on;

tcp_nodelay off;

keepalive_timeout 5;

gzip on;

gzipcomplevel 2;

gzip_proxied any;

gzip_types text/plain text/css application/x-javascript text/xml applic

ation/xml application/xml+rss text/javascript;

include /usr/local/nginx/sites-enabled/*;

}

Thanks for any tips,

Moses.

I just found my nginx error.log which had some interesting info in it

> 2009/12/16 11:47:09 [error] 12081#0: *1145 recv() failed (104: Connection reset by peer) while readi

ng response header from upstream, client: 221.133.197.254, server: www.domain.com, reques

t: "POST /wp-comments-post.php HTTP/1.1", upstream: "fastcgi://127.0.0.1:9000", host: "www.greenandg

oldrugby.com", referrer: "http://www.domain.com/ggr-moves-to-our- … al-server/">http://www.domain.com/ggr-moves-to-our-own-virtual-server/"

2009/12/16 11:47:10 [error] 12081#0: *1109 recv() failed (104: Connection reset by peer) while readi

ng response header from upstream, client: 203.33.167.125, server: www.domain.com, request

: "GET /wp-content/plugins/top-10/top-10-counter.js.php?toptenid=7401 HTTP/1.1", upstream: "fastcg

i://127.0.0.1:9000", host: "www.domain.com", referrer: "http://www.domain.com/

"

2009/12/16 11:47:11 [error] 12081#0: *1110 recv() failed (104: Connection reset by peer) while readi

ng response header from upstream, client: 203.33.167.125, server: www.domain.com, request

: "GET /wp-content/plugins/top-10/top-10-counter.js.php?toptenid=7294 HTTP/1.1", upstream: "fastcg

i://127.0.0.1:9000", host: "www.domain.com", referrer: "http://www.domain.com/

"

2009/12/16 11:50:15 [error] 12081#0: *1290 recv() failed (104: Connection reset by peer) while readi

ng response header from upstream, client: 150.101.187.176, server: www.domain.com, reques

t: "GET /feed/atom/ HTTP/1.1", upstream: "fastcgi://127.0.0.1:9000", host: "www.domain.co

m"

2009/12/16 11:55:12 [error] 12083#0: *1457 readv() failed (104: Connection reset by peer) while read

ing upstream, client: 71.235.154.48, server: www.domain.com, request: "GET /feed/ HTTP/1.

1", upstream: "fastcgi://127.0.0.1:9000", host: "www.domain.com"

and now, after a reboot, it's all working fine!

strange….

and now… nginx is again giving 502 errors.

Could it be related to a memory problem? Here are my charts from munin where the reboot just after midday on wednesday can be seen..

~~http://www.mosey.id.au/memory.png" />

my nginx error log is filling up again as well…

> //127.0.0.1:9000", host: "www..com", referrer: "http://www..com/"

2009/12/16 16:27:27 [error] 2369#0: *11692 recv() failed (104: Connection reset by peer) while reading response header from upstream, client: 203.14.52.45, server: www.***.com, request:

"GET /wp-content/plugins/top-10/top-10-counter.js.php?toptenid=7364 HTTP/1.0", upstream: "fastcgi:

//127.0.0.1:9000", host: "www..com", referrer: "http://www..com/"

2009/12/16 16:27:27 [error] 2369#0: *11690 recv() failed (104: Connection reset by peer) while reading response header from upstream, client: 203.14.52.45, server: www.***.com, request:

"GET /wp-content/plugins/top-10/top-10-counter.js.php?toptenid=7448 HTTP/1.0", upstream: "fastcgi:

//127.0.0.1:9000", host: "www..com", referrer: "http://www..com/"

2009/12/16 16:27:51 [error] 2369#0: *11720 readv() failed (104: Connection reset by peer) while reading upstream, client: 203.14.52.45, server: www.***.com, request: "GET /wallaby-tight-head-prop-of-the-decade/ HTTP/1.0", upstream: "fastcgi://127.0.0.1:9000", host: "www..com", referrer: "http://www..com/"

2009/12/16 16:27:53 [error] 2369#0: 11724 recv() failed (104: Connection reset by peer) while reading response header from upstream, client: 203.14.52.45, server: www.***.com, request: "GET /wp-content/plugins/top-10/top-10-addcount.js.php?toptenid=7367 HTTP/1.0", upstream: "fastcgi://127.0.0.1:9000", host: "www.**.com", referrer: "http://www.***.com/w

allaby-tight-head-prop-of-the-decade/"

2009/12/16 16:27:53 [error] 2369#0: *11724 recv() failed (104: Connection reset by peer) while reading response header from upstream, client: 203.14.52.45, server: www.***.com, request: "GET /index.php?akaction=akttcss HTTP/1.0", upstream: "fastcgi://127.0.0.1:9000", host: "www..com", referrer: "http://www..com/wallaby-tight-head-prop-of-the-decade/"~~

Moses,

Are you still having problems? What size linode are you using?

In your webmin boot up panel, do you have this:

php_cgi Yes Run php-cgi as app server

I swiftly read through your thread, saw you're using NGINX and followed some of theguv's instructions….

You're site is up right now, cached and loading great from Fort Worth, TX.

Cheers!

G'Day eyecool,

Thanks for the response!

I suppose I should have started a new thread when I blew away the CentOS instance and moved to Ubuntu Hardy.. There's too much in this thread now that's no longer relevant, and I don't expect people to read so much dribble!

Yep the site's up and flying, well at least it does for a few hours then certain things start giving nginx 502 errors. Things such as going to the wordpress admin page /wp-admin and leaving a comment on the blog.

The first time it happened I found a restart fixed it. The second time I was able to get it going again with

> /etc/init.d/php-fastcgi stop

/etc/init.d/php-fastcgi start

I have just re-read the Guv's guide and checked my system to find that I'm using libfcgi0ldbl for fcgi… and I suspect it's the cause of my systems' instability.

> dpkg -l | grep cgi

ii libfcgi0ldbl 2.4.0-7 Shared library of FastCGI

btw I'm on a Linode 360.

I had the same problem on both my linode and dedicated servers. The following script has kept both servers up and stable for months now: ~~[http://bash.cyberciti.biz/web-server/rhel-fedora-php-fastcgi-initd-script/" target="_blank">](http://bash.cyberciti.biz/web-server/rh … td-script/">http://bash.cyberciti.biz/web-server/rhel-fedora-php-fastcgi-initd-script/](

Also, it will appear in your bootup list in webmin as php_cgi. Make sure it is setup to start on boot. And make sure HTTP is set to not start at bootup.

Glad to see your wordpress site is humming along. I use wordpress mu on a quadcore, 6GB ram, SAS hd. I'm considering switching all to a linode. Testing the response to your site helped my confidence. I think I'd feel safer running a few linodes than my single point of failure dedicated server @ThePlanet.

I also found out that one of the lead wordpress developers and head wordpress mu guru has his site on a linode: http://ocaoimh.ie/

Cheers,

eyecool

Thanks again for the response.

I'm thinking of installing spawn-fcgi rather than the libfcgi0ldbl one from The Guv's guides.

I've gone without webmin after the move to Ubuntu, riding bareback in the console has been an experience! Is this what you're looking for:

> ps -ef | grep cgi

www-data 18031 1 0 21:07 ? 00:00:00 /usr/bin/php-cgi -q -b localhost:9000

www-data 18034 18031 2 21:07 ? 00:03:00 /usr/bin/php-cgi -q -b localhost:9000

www-data 18035 18031 3 21:07 ? 00:03:16 /usr/bin/php-cgi -q -b localhost:9000

www-data 18036 18031 2 21:07 ? 00:03:05 /usr/bin/php-cgi -q -b localhost:9000

www-data 18037 18031 2 21:07 ? 00:03:02 /usr/bin/php-cgi -q -b localhost:9000

www-data 18038 18031 3 21:07 ? 00:03:24 /usr/bin/php-cgi -q -b localhost:9000

root 28068 17473 0 22:53 pts/0 00:00:00 grep cgi

I haven't installed http, in fact the only part of apache I have got is the apache2-utils for the ab benchmarking.
> $ dpkg -l | grep http

$

Sounds like a good idea to move to a linode, alas I know far too little to give you any real advice at this point but I like the idea of a few nodes to load balance a busier site.

I don't use Apache so I can't tell what exactly is the problem. But I'm using Lighty and with that spawn-fcgi to spawn PHP processes. If I use single PHP process (children=0) whenever PHP burps, it crashes and the backend disappears.

If I use master + children processes (children = 1+), the children (worker) processes will not disappear in case of a crash. Also note the setting to restart PHP after X requests, afaik the default is 500 which might have caused your problem (restarts, but nothing is there to respawn new, if you used single PHP process).

Regarding those 500 requests, there are mixed opinions about it, but the common denominator is that PHP is/was bad at memory management and required process restart after certain number of requests, for example see this: http://redmine.lighttpd.net/wiki/1/Docs … nceFastCGI">http://redmine.lighttpd.net/wiki/1/Docs:PerformanceFastCGI

Well this is a weird one…

I installed spawn-fcgi and will monitor it to see how it runs.. strangely it doesn't like to startup after a reboot, is there any way on ubuntu where I can check if it's set to run?

I've created an init script at /etc/init.d/init-fastcgi

Running this script manually does start php-cgi
> ps -ef | grep php

www-data 2719 1 0 00:54 ? 00:00:00 /usr/bin/php-cgi -q -b localhost:9000

www-data 2722 2719 1 00:54 ? 00:00:08 /usr/bin/php-cgi -q -b localhost:9000

www-data 2723 2719 0 00:54 ? 00:00:04 /usr/bin/php-cgi -q -b localhost:9000

www-data 2724 2719 0 00:54 ? 00:00:03 /usr/bin/php-cgi -q -b localhost:9000

www-data 2725 2719 0 00:54 ? 00:00:03 /usr/bin/php-cgi -q -b localhost:9000

www-data 2726 2719 0 00:54 ? 00:00:02 /usr/bin/php-cgi -q -b localhost:9000

root 4120 2694 0 01:08 pts/0 00:00:00 grep php

I ran

update-rc.d init-fastcgi defaults

There is a symbolic link in the startup folder
> ls /etc/rc0.d -lthr | grep cgi

lrwxrwxrwx 1 root root 22 2009-12-17 00:39 K20init-fastcgi -> ../init.d/init-fastcgi

but I need to manually start it after a reboot. Any log I can check for this?

Cheers,

Moses.

I like /etc/rc.local for making stuff start at boot which isn't under the control of my package manager. I leave /etc/init.d alone, for the most part.

I basically followed this guide exactly for nginx with fastcgi php on nginx and everything worked fine:

http://tomasz.sterna.tv/2009/04/php-fas … on-ubuntu/">http://tomasz.sterna.tv/2009/04/php-fastcgi-with-nginx-on-ubuntu/

Thanks Jed for that startup link, it's actually working now.

I've switched over to spawn-fcgi (wish I'd read atambo's link first) and it's been a lot more stable… though I now get this message popping up in my console several times a minute!

> sh: host: not found

It doesn't seem to affect anything, but damn if I know where it's coming from!

Had something like that error on my install.

apt-get install host

If I recall, something in your site is trying to do reverse lookups on domains. Install 'host' and the error goes away. Or disable anything doing a reverse lookup.

@saiboogu:

Had something like that error on my install.

apt-get install host

If I recall, something in your site is trying to do reverse lookups on domains. Install 'host' and the error goes away. Or disable anything doing a reverse lookup.
Thanks saiboogu,

That fixed the host not found errors, and replaced them with "invalid query name 1" errors!

I googled this one and it led me to a conflict between that version of host and smf (simple machines forum). The solution was to install bind9-host, and the error has gone now so thanks for your help.

@Moses:

I googled this one and it led me to a conflict between that version of host and smf (simple machines forum). The solution was to install bind9-host, and the error has gone now so thanks for your help.

That explains why it was familiar, I'm also running SMF. :) Installing 'host' was enough to fix mine, guess we're dealing with different versions. Glad you could find it though.