is anyone familiar with nanny?
its starting to drive me nutty as i cant figure out what the problem is
:)
here's a sampling from my log :
Jan 24 15:46:05 happybox nanny[899]: running command
"/usr/sbin/ipvsadm" "-d" "-t" "63.89.64.206:80" "-r" "192.168.1.135"
Jan 24 15:46:05 happybox nanny[878]: making 192.168.1.1:80 available
Jan 24 15:46:05 happybox nanny[878]: running command
"/usr/sbin/ipvsadm" "-a" "-t" "63.89.64.204:80" "-r" "192.168.1.1" "-m"
"-w" "1"
Jan 24 15:46:07 happybox nanny[788]: making 192.168.1.154:80 available
Jan 24 15:46:07 happybox nanny[788]: running command
"/usr/sbin/ipvsadm" "-a" "-t" "63.89.64.196:80" "-r" "192.168.1.154"
"-m" "-w" "1"
Jan 24 15:46:10 happybox nanny[872]: shutting down 192.168.1.174:80 due
to connection failure
Jan 24 15:46:10 happybox nanny[872]: running command
"/usr/sbin/ipvsadm" "-d" "-t" "63.89.64.211:80" "-r" "192.168.1.174"
Jan 24 15:46:10 happybox nanny[787]: making 192.168.1.153:80 available
Jan 24 15:46:10 happybox nanny[787]: running command
"/usr/sbin/ipvsadm" "-a" "-t" "63.89.64.196:80" "-r" "192.168.1.153"
"-m" "-w" "1"
this is my system specs :
ultra 5
redhat 6.2
kernel 2.2.17
IP Virtual Server version 1.0.2 (size=4096)
nanny --version
Program Version: nanny 1.36
Built: 23/Jan/2001
A component of: piranha-0.4.17-7
i rolled all of this stuff from kernel source / piranha 0.4.17-7 sources
to make sure it was made properly for my red hat sparc 6.2...
anyway, i can ALWAYS hit the real servers from the load balancer... and
i never get timeouts or slow responses... but it seems everytime nanny
goes to do it, it never sees it as serviceable... but as soon as it goes
through the reentry time, it brings it back in service (just to repeat
the proccess all over again) ...
is there something i should be looking into? or is there some bug with
nanny? i used to have the load balancer running on a x86 platform, but
had misc problems few and far between as stated above....i assumed that
my machine was too slow and by moving to the ultra 5, i would be able to
bypass all of this... now i just get the problems quicker and more of
them :)
if it helps, this is a pretty latge nat based director.... it has about
20 real servers to manage across about 12 vips...
again, i can ALWAYS hit the real servers... so its not the real
servers.. i got that much squared away. :(
-ed
its starting to drive me nutty as i cant figure out what the problem is
:)
here's a sampling from my log :
Jan 24 15:46:05 happybox nanny[899]: running command
"/usr/sbin/ipvsadm" "-d" "-t" "63.89.64.206:80" "-r" "192.168.1.135"
Jan 24 15:46:05 happybox nanny[878]: making 192.168.1.1:80 available
Jan 24 15:46:05 happybox nanny[878]: running command
"/usr/sbin/ipvsadm" "-a" "-t" "63.89.64.204:80" "-r" "192.168.1.1" "-m"
"-w" "1"
Jan 24 15:46:07 happybox nanny[788]: making 192.168.1.154:80 available
Jan 24 15:46:07 happybox nanny[788]: running command
"/usr/sbin/ipvsadm" "-a" "-t" "63.89.64.196:80" "-r" "192.168.1.154"
"-m" "-w" "1"
Jan 24 15:46:10 happybox nanny[872]: shutting down 192.168.1.174:80 due
to connection failure
Jan 24 15:46:10 happybox nanny[872]: running command
"/usr/sbin/ipvsadm" "-d" "-t" "63.89.64.211:80" "-r" "192.168.1.174"
Jan 24 15:46:10 happybox nanny[787]: making 192.168.1.153:80 available
Jan 24 15:46:10 happybox nanny[787]: running command
"/usr/sbin/ipvsadm" "-a" "-t" "63.89.64.196:80" "-r" "192.168.1.153"
"-m" "-w" "1"
this is my system specs :
ultra 5
redhat 6.2
kernel 2.2.17
IP Virtual Server version 1.0.2 (size=4096)
nanny --version
Program Version: nanny 1.36
Built: 23/Jan/2001
A component of: piranha-0.4.17-7
i rolled all of this stuff from kernel source / piranha 0.4.17-7 sources
to make sure it was made properly for my red hat sparc 6.2...
anyway, i can ALWAYS hit the real servers from the load balancer... and
i never get timeouts or slow responses... but it seems everytime nanny
goes to do it, it never sees it as serviceable... but as soon as it goes
through the reentry time, it brings it back in service (just to repeat
the proccess all over again) ...
is there something i should be looking into? or is there some bug with
nanny? i used to have the load balancer running on a x86 platform, but
had misc problems few and far between as stated above....i assumed that
my machine was too slow and by moving to the ultra 5, i would be able to
bypass all of this... now i just get the problems quicker and more of
them :)
if it helps, this is a pretty latge nat based director.... it has about
20 real servers to manage across about 12 vips...
again, i can ALWAYS hit the real servers... so its not the real
servers.. i got that much squared away. :(
-ed