Finding the cause of Socket Errors

2012-02-12 Thread Sebastian Fohler
I get this socket error: haproxy[63881]: Server adserver.dev.adworxs.net/adserver2.dev is DOWN, reason: Socket error, check duration: 8ms. 1 active and 0 backup servers left. 0 sessions active, 0 requeued, 0 remaining in queue. The check duration changes with every error entry, and a manual

Re: Check backend servers

2012-02-12 Thread Sebastian Fohler
Thank you for all your help. I've fixed the problem now in configuring the backends to just answer with a index.php file as default, for every unspecific request. That seems to solve the problem for now. Greets Sebastian On 10.02.2012 14:12, Graeme Donaldson wrote: On 10 February 2012 14:50,

Re: Finding the cause of Socket Errors

2012-02-12 Thread Baptiste
On Sun, Feb 12, 2012 at 9:12 AM, Sebastian Fohler i...@far-galaxy.com wrote: I get this socket error: haproxy[63881]: Server adserver.dev.adworxs.net/adserver2.dev is DOWN, reason: Socket error, check duration: 8ms. 1 active and 0 backup servers left. 0 sessions active, 0 requeued, 0

Re: Finding the cause of Socket Errors

2012-02-12 Thread Sebastian Fohler
This is the definition of the backend checks: server adserver2 192.168.10.61:80 check inter 24 we ight 100 listen adserver.adworxs.net bind192.168.10.68:80 modehttp log global

Performance problems

2012-02-12 Thread Sebastian Fohler
I've checked the values Willy posted on the haproxy page. All my hardware configurations should meet the needs of haproxy. Still I have major performance problems. How do I best find out why? The logs tell me not nearly anything I neec to now to fix that problems. Since I use vm's to try

Re: Performance problems

2012-02-12 Thread Cyril Bonté
Hi Sebastian, Please, help us. You create too many threads at a time for too many different problems (or sometimes for the exact same issue), where most of the time the answer is in the documentation. It's really difficult to follow you. At this time, in this new thread, I can't guess which

Re: Performance problems

2012-02-12 Thread Łukasz Jagiełło
2012/2/12 Sebastian Fohler i...@far-galaxy.com: I've checked the values Willy posted on the haproxy page. All my hardware configurations should meet the needs of haproxy. Still I have major performance problems. How do I best find out why? The logs tell me not nearly anything I neec to now to

Re: Performance problems

2012-02-12 Thread Sebastian Fohler
The memory usage right now is around 65% now. So that shouldn't be the problem. We are talking about around 9 MBps Traffic, for around 700 session in short times for every webserver which are four right now. On 12.02.2012 12:24, Łukasz Jagiełło wrote: 2012/2/12 Sebastian

Re: Performance problems

2012-02-12 Thread Baptiste
2012/2/12 Łukasz Jagiełło jagiello.luk...@gmail.com: 2012/2/12 Sebastian Fohler i...@far-galaxy.com: I've checked the values Willy posted on the haproxy page. All my hardware configurations should meet the needs of haproxy. Still I have major performance problems. How do I best find out why?

Re: Performance problems

2012-02-12 Thread Sebastian Fohler
Sorry I thought it's a completly different thread since it does not address the check problems at all. But if it would have been better to leave all those questions in one thread, I apologize. The values I'm talking about are the performance examples to be found on the haproxy website, so not

Re: Performance problems

2012-02-12 Thread Baptiste
Hi Sebastian, Can you take a screenshot of your stats page and share it with us? Then, please do the changes below in your configuration: turn nbproc to 1 increase the maxconn on the frontend to 4000 cheers

Re: Performance problems

2012-02-12 Thread Sebastian Fohler
Hi, On 12.02.2012 12:54, Baptiste wrote: lukaszatjagiellodotorg Hi, When speaking about virtualization, there are so many factors to take into account. - which hypervisor are we speaking about ? VMWare ESXi 5.0 - what phisical NIC are using? Broadcom Corporation Broadcom NetXtreme II

Re: Performance problems

2012-02-12 Thread Sebastian Fohler
I did sent you the screenshot, before the changes. Just to clear that out. How do I know that HAProxy is taking all the CPU resources? PFSense shows the CPU and NIC Traffic resources, I get a rising CPU load until it's 100% and after that haproxy stops the adserver.adworxs.net frontend the

Re: Performance problems

2012-02-12 Thread Cyril Bonté
Le 12/02/2012 13:21, Sebastian Fohler a écrit : Hi, On 12.02.2012 12:54, Baptiste wrote: lukaszatjagiellodotorg Hi, When speaking about virtualization, there are so many factors to take into account. - which hypervisor are we speaking about ? VMWare ESXi 5.0 Take into account that

Re: Performance problems

2012-02-12 Thread Sebastian Fohler
On 12.02.2012 14:24, Cyril Bonté wrote: Take into account that virtualization can be a bottleneck for a network component like haproxy. I know that, that's the reason I used one physical nic especially for the haproxy instances. Sure there could still be some problem, but that would be a

Re: Performance problems

2012-02-12 Thread Baptiste
Your nbproc is still 2. You should turn it to 1. You should also remove the option httpclose, replacing it by option http-server-close. It will save some packets on the client side. Are the backend servers VMs too? Are they hosted on the same vsphere server? How many physical network cards are

Re: Performance problems

2012-02-12 Thread Sebastian Fohler
I've changed nbproc to 1 now, as I did change the httpclose option, it seems to help a little. The adserver.adworxs.net backends are physical servers. The adserver.dev.adworxs.net backends are vm's. There are four physical NICs used by the vsphere server. Greets, Sebastian On 12.02.2012 14:42,

Re: Performance problems

2012-02-12 Thread Baptiste
On the last screenshot you set, you had a lot of retries... Which means there may be some issues for your HAProxy to reach the server sometimes. Have you checked all the network settings (speed and duplex) on both servers (web and vsphere) and switches?

Re: Performance problems

2012-02-12 Thread Sebastian Fohler
I've noticed that too. The problem is, when I try to reach the backendservers themselves (they are all reachable by there own name, adserve1/adserver2/...). The don't show any problems at all, the question in that case is, how do I find out which error they throw, seen from the lb end. Sure I

Re: Performance problems

2012-02-12 Thread Cyril Bonté
Le 12/02/2012 22:31, Sebastian Fohler a écrit : I've noticed that too. The problem is, when I try to reach the backendservers themselves (they are all reachable by there own name, adserve1/adserver2/...). The don't show any problems at all, the question in that case is, how do I find out which

Re: Log 400 bad request

2012-02-12 Thread Willy Tarreau
Hi Sander, On Fri, Feb 10, 2012 at 04:28:39PM +0100, Sander Klein wrote: Hi All, I'm having a small problem with non RFC2616 requests. I would like to log them, but haproxy only logs: cluster1-in cluster1-in/NOSRV -1/-1/-1/-1/0 400 1951 - - PR-- 235/235/0/0/0 0/0 {|||} {} BADREQ Is

Re: Finding the cause of Socket Errors

2012-02-12 Thread Willy Tarreau
On Mon, Feb 13, 2012 at 08:01:11AM +0100, Willy Tarreau wrote: On Sun, Feb 12, 2012 at 09:12:47AM +0100, Sebastian Fohler wrote: I get this socket error: haproxy[63881]: Server adserver.dev.adworxs.net/adserver2.dev is DOWN, reason: Socket error, check duration: 8ms. 1 active and 0