Re: Limit to number of items in an ACL matching src IP
Hi Andrew Thanks for this. I was s close, I looked in acl.c and didn't see anything there. Graeme "not a C hacker at all" Donaldson. :-) On 18 February 2010 14:34, Andrew Azarov wrote: > Hi Graeme, > > ./include/common/defaults.h > > #define LINESIZE 2048 > > #define MAX_LINE_ARGS 64 > > But you can change them any time - repercussions: memory and cpu usage. > Mine are > > #define LINESIZE 65536 > > #define MAX_LINE_ARGS 4096 > > everything runs fine. > > BRG, > Andrew > > > On 16.02.2010 10:28, Graeme Donaldson wrote: > >> Hi >> >> I'm playing around with something like this: >> >> acl src_goaway src 10.0.0.1 >> redirect location http://example.com/goaway.html >> >> I have seen examples in the docs where src is specified as multiple IPs in >> a single ACL, but I don't see any mention of how many IPs can be in a single >> ACL. >> >> Did I miss this in the docs? >> >> Thanks, >> Graeme. >> >> >
Re: Backup servers in 1.3
Hi Chris, Le Jeudi 18 Février 2010 15:57:11, Chris Sarginson a écrit : > Hi, > > Earlier today we took all our primary webservers offline to perform some > maintenance on our webservers. There is one backup server configured in > haproxy which simply hosts a single index page, and rewrites all URLS to > display this. > > Following bringing the servers back online, traffic was still being > directed to the backup server, so my question is thus: > > What is the correct behaviour for a server configured as a backup in > haproxy, once the normal servers are back online? It depends on the configuration. From your configuration, the behaviour is correct because of the "sorry" cookie : as soon as a response provides this cookie, all the next requests providing the same value will be directed to this server. If you want the requests return to the servers as soon as they're up again, you can remove "cookie sorry" on the backup server. Then this request will be load balanced to one of the active servers. > server sorry 10.10.10.184:80 cookie sorry check inter 3000 rise 2 fall 2 > weight 1 backup could become : server sorry 10.10.10.184:80 check inter 3000 rise 2 fall 2 weight 1 backup -- Cyril Bonté
Re: Enabling logging on solaris
On Thu, Feb 18, 2010 at 2:09 PM, Mallin, Eoin wrote: > > Hi all, > I'm trying to get HAProxy logging enabled on solaris 9. My HAProxy > configuration file looks as follows: > > log 127.0.0.1 local0 info > > /etc/syslog.conf now has the following entry > > local0.* /haproxy/log/haproxy.log > > Is there another step to allow syslog accept socket connections? > > Many thanks, > Eoin > > > Eoin Mallin > __ > ITS Architecture > Castle View > Georges Street > Dublin 2 > VPN: 35379 > > Just a hint, try to see if syslog is listening on udp port 514, netstat or, if its installed, lsof -i :514 can help you on that. -- Germán Gutiérrez OLX Operation Center OLX Inc. Buenos Aires - Argentina Delivering common sense since 1969 . The Nature is not amiable; It treats impartially to all the things. The wise person is not amiable; He treats all people impartially. No afecta al sitio, no necesita QA.
Enabling logging on solaris
Hi all, I'm trying to get HAProxy logging enabled on solaris 9. My HAProxy configuration file looks as follows: log 127.0.0.1 local0 info /etc/syslog.conf now has the following entry local0.* /haproxy/log/haproxy.log Is there another step to allow syslog accept socket connections? Many thanks, Eoin Eoin Mallin __ ITS Architecture Castle View Georges Street Dublin 2 VPN: 35379 * Please note that Revenue cannot guarantee that any personal and sensitive data, sent in plain text via standard email, is fully secure. Customers who choose to use this channel are deemed to have accepted any risk involved. The alternative communication methods offered by Revenue include standard post and the option to register for our (encrypted) secure email service. http://www.revenue.ie/en/practitioner/secure-email.html *
Re: how to enable syslogging
On Wed, Feb 17, 2010 at 11:02 PM, Nelson Serafica wrote: > I need to enable syslog of haproxy. I search in Google but can't find the > right one. I'm using Fedora 8 so syslog was rsyslog. I edit > /etc/rsyslog.conf and put the ff: > > local0.* /var/log/haproxy.log > local1.* /var/log/haproxy-1.log > > Then I edit /etc/sysconfig/rsyslog and put "SYSLOGD_OPTIONS="-m 0 -r". After > that I restart rsyslog service. I see the file haproxy.log and haproxy-1.log > but when I restart the haproxy, nothing comes up in the log file. I even > browse the site but still no input on the log file. Am I missing something? > Most of the suggestion was on the above but its not working on mine. > > Does anyone also experience the same problem and found a solution? > I had trouble identifying this on debian lenny, we need rsyslog to listen on the 514 UDP port in my case I needed to uncomment the following lines on the rsyslog.conf file: #$ModLoad imudp #$UDPServerRun 514 That's because the parameter that enables the UDP listening is now deprecated. Hope this helps. Regards, -- Germán Gutiérrez OLX Operation Center OLX Inc. Buenos Aires - Argentina Delivering common sense since 1969 . The Nature is not amiable; It treats impartially to all the things. The wise person is not amiable; He treats all people impartially. No afecta al sitio, no necesita QA.
Backup servers in 1.3
Hi, Earlier today we took all our primary webservers offline to perform some maintenance on our webservers. There is one backup server configured in haproxy which simply hosts a single index page, and rewrites all URLS to display this. Following bringing the servers back online, traffic was still being directed to the backup server, so my question is thus: What is the correct behaviour for a server configured as a backup in haproxy, once the normal servers are back online? Cheers Chris Config as follows (hostnames changed to protect the innocent ;): global log 127.0.0.1 local3 #log loghostlocal0 info maxconn 4 chroot /var/lib/haproxy user haproxy group haproxy daemon #debug #quiet stats socket /tmp/stats defaults log global modehttp option httplog option dontlognull retries 3 option redispatch maxconn 5000 contimeout 5000 clitimeout 5 srvtimeout 5 option httpchk option httpclose option forwardfor # enable web-stats at /haproxy?stats listen 3632-10.10.10.172:80 10.10.10.172:80 mode http default_backend 3632_http backend 3632_http balance roundrobin mode http cookie SERVERID insert nocache indirect option httpchk HEAD /checklb.php HTTP/1.1\r\nHost:blah.test.co.uk contimeout 5000 srvtimeout 5 server web01 10.10.10.193:80 cookie web01 check inter 3000 rise 2 fall 2 weight 1 server web02 10.10.10.192:80 cookie web02 check inter 3000 rise 2 fall 2 weight 1 server web03 10.10.10.191:80 cookie web03 check inter 3000 rise 2 fall 2 weight 1 server sorry 10.10.10.184:80 cookie sorry check inter 3000 rise 2 fall 2 weight 1 backup
Re: Limit to number of items in an ACL matching src IP
Hi Graeme, ./include/common/defaults.h #define LINESIZE 2048 #define MAX_LINE_ARGS 64 But you can change them any time - repercussions: memory and cpu usage. Mine are #define LINESIZE 65536 #define MAX_LINE_ARGS 4096 everything runs fine. BRG, Andrew On 16.02.2010 10:28, Graeme Donaldson wrote: Hi I'm playing around with something like this: acl src_goaway src 10.0.0.1 redirect location http://example.com/goaway.html I have seen examples in the docs where src is specified as multiple IPs in a single ACL, but I don't see any mention of how many IPs can be in a single ACL. Did I miss this in the docs? Thanks, Graeme.
RDP Cookie Persistence
We have been using RDP cookie persistence and noticed that sometimes the distribution of the connections is not exactly even. The problem we suspect is using round robin do distribute the load you can end up with strange loadings where people have disconnected in groups, is it possible to use least connected for the balancing algorithm. The problem is if you have 2 terminal servers server a, server b, two people connect to a then two people connect to b ( so connecting, a,b,a,b) the next person to come along connects to a again. then all the people leave a so there are no people connected to a, and 2 people connected to b the next connection will go to b so we have 3 on b and 0 on a, obviously this can be expanded then you end up with strange loadings. Hopefully I have explained that correctly Any thoughts would be greatly appreciated Mark
american consumer database {300,000 records all with emails}
I have many good quality lists Doctors and Dentists & more at decent prices. send and email to: jef...@accuratedata.co.cc I'll get you all the details and samples. For your subscription status modification please contact rem...@accuratedata.co.cc
