Load Balancer From A Linux Box
Apr 15, 2008I wish to build a load balancer on a linux box that will load balance http/https/smtp/pop3 and ftp
what technologies i need to look at?
I wish to build a load balancer on a linux box that will load balance http/https/smtp/pop3 and ftp
what technologies i need to look at?
Im currently looking around for a Load Balancer within the budget of about £2,500
Anyone have any suggestions?
Will that budget get me one that will do the job or would I be wasting my money and should maybe save up some more cash?
Setup:
2x Web Servers
1x DB Server
1x Storage / Media Servers
*Expanding to 3/4x Web Servers, 2x DB Servers, 2x Storage / Media Servers*
I have been looking at:
F5 BIG IP D25 Load Balancer
[url]
and
F5 BIG IP D30 Load Balancer
[url]
But im unsure on what the difference between these are? as I cant seem to find there specs and what they can / cant do..
Anyone know of a good load balancer for Linux that is able to not only distribute load evenly amongst many servers, but one that's able to allocate servers based on the resource?
For instance:
I want www.mysite.com/dogs to be load-balanced between servers A, B, and C.
I want www.mysite.com/cats to be load-balanced between servers X, Y, and C.
Is there any such solution available?
Something with a failover load balancer that you can assign a static IP address to and it'll sit in front of a number of lowend virtual private servers.
I have two servers that I run in Los Angeles (used to host gaming servers for a small community I run) for personal use that seem to be a little jumpy with server performance.
Would a load balancer help with this situation?
And for the n00b question, what exactly is a load balancer.
Very new to hosting servers, so I appreciate a gentle response
I'm using Virtuozzo on two server.
1) Can I do this with only ONE Load Balancer?
I wish I could have many "LB", like one for the WWW serveur, an other one for the DB serveur, my "personnal" VPS without LB...
i.e. :
WWW -> SRV1 SRV2
DB1 -> DB1 DB2
EXCESSNET -> EXCESSNET1
Do I need 3 LB to do this or I can do it with only one? The screenshot of Barracuda LB seem to show many "virtual LB", so I guest it's a yes?
2) So... it's service only?
On the Barracuda and some page, they are talking about "service" and not "server" LB. Can I LB a whole serveur ? Like all service from an IP xxx.xxx.xxx.234 to a server ? Or I can only do "HTTP -> SRV1" "DNS -> SRV1"? If no, what happen to service that the LB didn't know (like cPanel)?
3) An Internal Network? what About cPanel?
According to my search, LB are working like this:
EXTERNAL IP of SERVICE (xxx.xxx.xxx.200) -> INTERNAL IP SERVER 1(10.0.0.25), INTERNAL IP SERVER 2 (10.0.0.26)
So, I need to configure the server with an Internal IP cause else, it will conflit on the "local network". The probleme is, cPanel will be using the "Internal IP" to make all of his configuration... so if a user add a new website, he will configure it with 10.0.0.25, this won't since the external IP will be xxx.xxx.xxx.200 and Apache won't anwser to this virtual host. What can I do?
Can I do this ? (same internal IP as the external?)
EXTERNAL IP (70.234.125.123) -> SERVER 1 INTERNAL IP (70.234.125.123), SERVER 2 INTERNAL IP (10.0.0.26)
If yes, I will configure the second serveur without cPanel and only sync the configuration and files. So in case of "crash" of the server 1, cPanel won't be working, but website won't be down...
We are planning for a clustering archirecture for our mail servers,The basic idea is put all of mailservers behind a load balancer which will monitor and distribute the n/w load as server load and forward the requests accordingly can u suggest any good hardware loadbalancer which could give us 'server load balancing' as well as n/w load balacing.
I would also like to know if it is a good idea to go for a software load balancer(like linux heartbeat) or to h/w load balancer.
traffic has been surging lately and is expected to continue surging. i am already on the fastest chipset my webhost has.
current configuration
single server (centos), webserver (apache), single filesystem. many dynamic pages (perl script) but no mysql database. no php.
ideal configuration
multiple webservers (apache), single file system.
assuming my scripts do not write to the filesystem (read only), what do i need to install or do in order to send a request for www.domain.com to server1.domain.com or server2.domain.com or server3.domain.com according to the load of each? it should look the same at the front end. anyone know what i need to get?
if it is possible to hide the Barracuda Load Balancer signature in headers ?
If you check here : [url]
or do a telnet 80 to that same FQDN and send "HEAD / HTTP/1.0" + "RETURN" you will be able to see in the headers the server signature (BarracudaHTTP 1.00).
Is there any way to hide this?
I'm trying to find a solution but without any result so far,
This is the story:
We have too server in Israel, in one server we have:
Win 2003 STR that use for IIS (IBM RAID 5)
And in the second we have:
Win 2003 STR + SQL SERVER 2005 workgroup.
After 1 year running in Israel we like to go out to the world with our products.
This is what I think we need to do but we still don't figure how to that:
1. Still use the same server in Israel.
2. Buy more server outside Israel, Lets say UK and set a Load Balancer between the server in UK to Israel (both IIS and SQL).
3. In the Load balancer we like to set that if the user come from europe you will go to the server in UK and if he come from Israel you go the server there.
4. If one of the server we be go down all the the users will sends to the server outside there country.
5. In the future we like to add more server in the world on the same setting.
What do you say? how can I do something like that?
What company are dealing stuff like that?
Anyone know of a hardware load balancer that can load balance my apache and mysql using one device? I mean right now I can use the ultra monkey for the load balancer the cluster mysql. But I want to go hardware.
I was thinking of F5 BIGIP? will that work? anyone know Please give me some choices..
I have a setup where Apache connects to a F5 load balancer which in turn balances between two jboss app servers.
Apache using mod_proxy -> F5 (hardware load balancer) -> 2 jboss application server
It uses jsessionid. I sometimes get 500 errors for the post methods. I think the request goes to the incorrect jboss server because of the F5 load balancer. Everything works just right when I shutdown one of the jboss app servers.
I am thinking of using a Load Balancer + serveral VPSs solution.
Is there any good recommendation ? It would be great if the hosting company can provide the load banlancer for free.
I am using single Apache HTTP Server (2.2.23) as a Load Balancer with two IBM Websphere application server nodes (other machines). I have deployed the simple text based helloWorld application and it works fine with load balancer. But When I deploy the real application that contains images,css file , java script file. It loads the page without images and show me simple text and gives me the following Exception on error_logs and similar kind of exceptions
[error] [client 192.217.71.77] File does not exist: /usr/local/apache2/htdocs/application, referer: http://192.168.141.17/application/faces/test.jsp
Interestingly, when I access the application without load balancer, it also works fine.
Here are the load balancer settings
A little bit excited as the Barracuda is coming soon...
We have 1 forum server, and it's being access internally by all the classrooms, students, teachers, etc. How should I approach this setup? The current hostname for the forum machine is forum... how should I setup the cluster? forum2?
I have a short question concerning Mod Balancer and Stickiness:
At the moment I am achieving stickiness with a cookie.
Header add Set-Cookie "ROUTEID=.%{BALANCER_WORKER_ROUTE}e; path=/" env=BALANCER_ROUTE_CHANGED
However, I am getting complaints by users which do not allow coockies in their browser settings that my page does not work properly.
Is there any way to achieve Stickiness with out cookies (e.g. add the information to the URL or something like that)?
I'm using mod_proxy_balancer in Apache 2.2.29 with OpenSSL on Windows to distribute HTTPS requests to two other back end servers, both of which are also Apache 2.2.29 with OpenSSL on Windows.
DumpIO shows that the requests are being decrypted on the balancer, and then sent unencrypted to the two workers. I would like to have the requests pass through the balancer without decrypting there, and let the two back end worker servers do the decryption. Is that possible? If so, how?
When I tried simply disabling SSL on the balancer, the requests fail, with nothing logged at all.
This is a RHEL3 box. It has a couple hundred sites, most of which are pretty quiet most of the time, lately I can't get the load below 3 most of the time. I've tried all the usual TOP PS and the rest, but I'm not seeing my problem. I've attached a text file with extended outputs. but here's what I'm seeing now from TOP and VMSTAT. I'm not sure what else I could be looking at.
I've been watching, and it seems to be Apache most of the time at the top of TOP, and I'm logging slow-queries for myslq, but there aren't many there.
top
10:02:03 up 3 days, 21:23, 2 users, load average: 2.81, 2.83, 3.05
140 processes: 137 sleeping, 2 running, 1 zombie, 0 stopped ....
the Load averages in Linux I tend to find confusing, I've read that its the system load averaged over 1, 5 and 15 minutes but beyond that I never seem to find a clear cut answer as to what is considered a normal system load.
For instance I have a p4 D 3Ghz with 2gb ram and a single 250gb SATA HDD
the 15min average sits around 0.85 during the busiest time Is that good bad or indifferent?
The server hosts my sites and that of a few friends.
[url]shows what MRTG is outputting about the load averages.
It also says the CPU is 520% although I believe it multiplies it by 10? (i changed the title to say that) so its around 52%
I fixed that there is a process that loads the server and much why it does not.
Name process: md2_resync
I've been having a few issues with the control panel navigation is running slow even though I have heaps of resources available and free. The server is located locally and have no problems surfing websites attached to the same server.
When I login as admin, it can take 25 seconds to load into the first screen.
CPU GenuineIntel, Intel(R) Xeon(R)CPU E5649 @ 2.53GHz
Version Parallels Plesk v12.0.18_build1200140606.15 os_CentOS 6
OS CentOS 6.6 (Final)
Memory Available 6GB used 2.5-3GB
HDD: 300GB/ approx 130GB Free
I have tried issuing a /etc/init.d/psa restart command and has made no difference.
Plesk Backup work when we make instant backup but not with scheduled backup.
There is nothing in the log files (pmmcli.log & migration.log).
How can we correct?
Plesk 12 / Centos
Plesk Panel 12.0.18 , cant load pages in controlpanel (web host edition) it is making timeout all the time and also cannot send mail (roundcube).
What have parallels done to the a nice panel!
I have a problem with my plesk panel. Ever since I rebooted my server it won't load the IP:8443 page at all.
View 5 Replies View RelatedI am running Plesk Version: 12.0.18 on CentOS.Earlier today my service provider had a massive issue and my server was down for several hours. Upon it being back online i've restarted but now I cannot load any of my pages. I have 10 domains setup and if I ping their name servers or the server IP i get packets received now.
One of my installed domains is [URL] ... now if you try and load this url you get a time out but if you try to access admin page URL....None of the subscriptions are able to load their websites when i type in their URL and well it worked before it went down.
I am troubleshooting a problem where the php5 package was replaced and caused lots of psa/plesk packages to be removed. This system details are;
# cat core.version
11.5.30 Ubuntu 12.04 115140407.17
# lsb_release -a
Description: Ubuntu 12.04.2 LTS
Codename: precise
# uname -a
Linux my.host.name 3.2.0-45-generic #70-Ubuntu SMP Wed May 29 20:12:06 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux
The plesk panel is up, the websites are running, and the email services are running, but there are many errors like so;
==> error.log <==
PHP Warning: PHP Startup: Unable to load dynamic library '/usr/lib/php5/20090626/json.so' - /usr/lib/php5/20090626/json.so: cannot open shared object file: No such file or directory in Unknown on line 0
I also get the following error when inspecting the json module:
root@vps:/etc/php5/mods-available# php -i|grep -i json
PHP Warning: PHP Startup: Unable to load dynamic library '/usr/lib/php5/20090626/json.so' - /usr/lib/php5/20090626/json.so: cannot open shared object file: No such file or directory in Unknown on line 0
/etc/php5/cli/conf.d/20-json.ini,
[Code] ....
What I need to install to get that json.so file back?
I am creating new sites in my panel and they are only loading the default page.
The IP is pointed correctly or I would not get the default plesk page.
All the other service seem to work ok. I can ftp in and I can edit files with the online file manager.
If I use the preview feature I can see the changes I make to the files, but that is the only way.
I looked in /var/www/vhosts/system/ and the conf files are present.
What else can I check to see why the sites are not working?
Today, without any change on the server side my single running website on my server started running low. After checking the running processes I realised I have these 8 processes running with high cpu load something that didn't had before.
4245 user 20 0 283m 103m 10m R 85.1 0.3 6:44.88 /usr/bin/php-cgi -c /var/www/vhosts/system/domain/etc/php.ini
4243 user 20 0 241m 68m 8736 R 83.8 0.2 6:33.45 /usr/bin/php-cgi -c /var/www/vhosts/system/domain/etc/php.ini
4259 user 20 0 254m 82m 8720 R 83.5 0.3 6:11.00 /usr/bin/php-cgi -c /var/www/vhosts/system/domain/etc/php.ini
4244 user 20 0 282m 103m 10m R 82.1 0.3 6:46.49 /usr/bin/php-cgi -c /var/www/vhosts/system/domain/etc/php.ini
4247 user 20 0 248m 76m 8712 R 78.8 0.2 6:50.31 /usr/bin/php-cgi -c /var/www/vhosts/system/domain/etc/php.ini
4264 user 20 0 258m 85m 8712 R 72.9 0.3 6:49.83 /usr/bin/php-cgi -c /var/www/vhosts/system/domain/etc/php.ini
apache error log doesn't show any warnings, errors.
PRODUCT, VERSION, OPERATING SYSTEM, ARCHITECTURE
Parallels Plesk Panel 11.5.30 MU #44 CentOS 6.5 (Final) 64bit
2 of our our servers are suddenly experiencing high Wait I/O Times, and high Load Averages during the backup process. During this period the Plesk grinds to a halt, sometimes crashing out completely (although SSh is still possible. We have been in talks with our server suppliers (assuming this would be node related) however they have done a lot of testing etc. and categorically state the node is fine , with no other users affecting it.
STEPS TO REPRODUCE : We backup the server using the scheduled backup service and Wait I/O immediately goes up.
ACTUAL RESULT: Plesk downtime / Website downtime
EXPECTED RESULT: No downtime, successful back up
Some other info :
All other processes (MYSQL, apache, Nginx etc) are all running between 1 - 10%
Partition "/usr" utilization 4.2% used (1.81 GB of 43.3 GB) (?)
Partition "/var" utilization 50.6% used (61.8 GB of 122 GB) (?)
We are struggling to identify what has changed on the server that would cause this sudden change.
VPS hosting, stats in my sig below. Have poor performance with a drupal and civicrm(drupal based) website and looking to troubleshoot.
Main issue is pages take too long to load, sometimes over 5 seconds and more! Below is my.cnf and MYSQLtuner results.
So my.cnf
Code:
[mysqld]
set-variable=local-infile=0
datadir=/var/lib/mysql
socket=/var/lib/mysql/mysql.sock
user=mysql
[Code] .....
PRODUCT, VERSION, MICROUPDATE, OPERATING SYSTEM, ARCHITECTURE
Parallel Plesk 11,0,9, update 63; OS centOS 6.
PROBLEM DESCRIPTION
when I try to access a domain to make changes, the system gives me this error
Message :
Unable to load object of type Account with id = 18: Account: unable to select: no such row in the table
file common_func.php3
line 3327
Type PleskFatalException
STEPS TO REPRODUCE
In ADMINISTRATION panel, clik DOMAIN -> click on one of 9 domains (only one don’t work)
ACTUAL RESULT
Error message
EXPECTED RESULT
normal operation of domain