Hi Willy, I catch a new big problem about the Haproxy after it runs two days successfully I add a command in my config file
reqirep ^([^\ ]*)\ /loot/(.*) \1\ /bigloot/tutorial/index/\2 it replace the path /loot/att to /bigloot/tutorial/index/att. Then the Haproxy hangs every hours and I get the error as follows: Nov 17 23:11:31 239488-web3 kernel: haproxy[11343]: segfault at 00000000203ab000 rip 0000003ef9c7c51a rsp 00007fff7fa050a8 error 6 I google and can not find any answer for that. Here is some information about my haproxy. # haproxy -vv HA-Proxy version 1.4.9 2010/10/28 Copyright 2000-2010 Willy Tarreau <[email protected]> Build options : TARGET = linux26 CPU = generic CC = gcc CFLAGS = -m64 -march=x86-64 -O2 -g OPTIONS = Default settings : maxconn = 2000, bufsize = 16384, maxrewrite = 8192, maxpollevents = 200 Encrypted password support via crypt(3): yes Available polling systems : sepoll : pref=400, test result OK epoll : pref=300, test result OK poll : pref=200, test result OK select : pref=150, test result OK Total: 4 (4 usable), will use sepoll. The configure file is the same that we discuss before, I just post it here: ========================================================== global maxconn 32000 # Total Max Connections. This is dependent on ulimit daemon #nbproc 1 # Number of processing cores. Dual Dual-core Opteron is 4 cores for example. pidfile /var/run/haproxy.pid #stats maxconn 5 stats socket /var/run/haproxy.sock mode 600 level admin stats timeout 1m defaults mode http clitimeout 5000 srvtimeout 32000 contimeout 10000 option http-server-close maxconn 30000 log global # log 127.0.0.1 local3 log 127.0.0.1 local3 err option httplog listen http_proxy bind 192.168.100.44:80 balance roundrobin # Load Balancing algorithm option httpchk HEAD /wap2/login.jsp HTTP/1.0 option forwardfor # This sets X-Forwarded-For reqirep ^([^\ ]*)\ /loot/(.*) \1\ /bigloot/tutorial/index/\2 stats enable stats realm Haproxy stats uri /admin stats auth test:test stats refresh 30s stats hide-version #option httpclose option http-server-close capture cookie JSESSIONID len 52 #stick-table type string len 52 size 2g expire 1h cookie ICEB_HTTP_SERVERID insert nocache indirect #appsession JSESSIONID len 52 timeout 3h mode query-string appsession JSESSIONID len 34 timeout 11m ## Define your servers to balance server server1 192.168.100.64:8280 weight 2 maxconn 7500 check inter 2000 rise 2 fall 3 cookie A server server2 192.168.100.65:8280 weight 3 maxconn 7500 check inter 2000 rise 2 fall 3 cookie B server server3 192.168.100.66:8280 weight 3 maxconn 7500 check inter 2000 rise 2 fall 3 cookie C server server4 192.168.100.39:8280 weight 2 maxconn 7500 check inter 2000 rise 2 fall 3 cookie D ========================================================== do you know what I can do for that? does the reqirep problem? Best Regards, Hogan On Wed, Nov 17, 2010 at 10:52 PM, Hogan Yu <[email protected]>wrote: > Hi Willy, > You are right, I misunderstand the log and thanks for your patient. > > Best Regards, > > Hogan > > > On Tue, Nov 16, 2010 at 3:03 PM, Willy Tarreau <[email protected]> wrote: > >> On Tue, Nov 16, 2010 at 09:03:17AM +0800, Hogan Yu wrote: >> > Hi Willy, >> > Sorry for reply so late, I test my configuration according your >> > suggestion. It is correct that We need stunnel in front of haproxy to >> > decipher HTTPS. I misunderstand the configuration on https and it works. >> We >> > have lots of long connection on our application server which use GWT to >> do >> > the chat service with comet support, So I need to change the srvtimeout >> to >> > about 32s. >> > We are a WAP website that use access our website by mobile phone >> browser, >> > there are about one-third users browser can not support Cookie >> according >> > to our access log. When they first access the Haproxy ,the session >> status >> > is --NI and when they come back again, the status change to --VN. >> > ( >> > >> > The third character means: >> > >> > N : the client provided NO cookie. This is usually the case for new >> > visitors, so counting the number of occurrences of this flag >> in the >> > logs generally indicate a valid trend for the site >> frequentation. >> > >> > >> > V : the client provided a VALID cookie, and was sent to the associated >> > server. >> > >> > The last character means: >> > >> > >> > N : NO cookie was provided by the server, and none was inserted either. >> > >> > I : no cookie was provided by the server, and the proxy INSERTED one. >> > Note that in "cookie insert" mode, if the server provides a >> cookie, >> > it will still be overwritten and reported as "I" here. >> > >> > ) >> > >> > >> > The VN means that the browser don't support cookie but Haproxy already >> get >> > the mapping of jsessionid and server, so it still be transfer the >> request to >> > correct backend server. >> >> No, it's the opposite, VN means that the browser did correctly present >> the cookie it learned at the previous request. The first request shows >> NI (No cookie presented, one cookie Inserted) and subsequent requests >> show VN (Valid cookie presented, No cookie inserted). So for these users >> that means that you don't need the appsession. >> >> The correct way to check for cookie support is to grep for requests >> showing >> the JSESSIONID in the logs and check if any of them has the NI flags. Only >> these ones are interesting to check because JSESSIONID means that the >> browser >> has already seen the cookie and NI means that it did not present it, which >> clearly indicates a lack of cookie support. And I'd bet there are between >> none at all and very few. >> >> Regards, >> Willy >> >> > >
