OVH Community, your new community space.

OCO service error


cioby23
08-02-2011, 19:00
Hi,
I managed to get OCO service to work. It appeared that the http script file did not have execute permissions set. After setting them all went fine.
Thanks for the tip.
Regards,

marks
07-02-2011, 09:49
for some reason, the OcO service that checks the state of the server fails, and doesn't write the updated state of the server.

It could fail for several reasons like non-writable folder, some of the network checks fail, apache not running (remember that by default only checks the port 80, ...).

I suggest you to follow the scripts that OcO runs (bin folder) and see in what step is the service writing "398 - Result too old"

cioby23
07-02-2011, 08:56
I have allowed ping to the load balancing IP and it works ok for around 5 minutes until the OCO service fails with the OCO Probe: 398 - Result too old error. If I modify the http file from the result folder on both servers and put it on KO(300) and then OK(200) again it works for 5 minutes.
That's weird.
Currently the interfaces on the http script from /usr/local/oco/bin/60sec are set like below:

#-- Get info about server
15 my $HOST=`/sbin/ifconfig lo | grep inet`;
16 $HOST=~ m/.*r\d+\.\d+\.\d+\.\d+).*/;
17 $HOST=$1;
18
19 unless ($HOST)
20 {
21 $HOST=`/sbin/ifconfig eth0:0 | grep inet`;
22 $HOST=~ m/.*r\d+\.\d+\.\d+\.\d+).*/;
23 $HOST=$1;
24 }

I'm doing something wrong here ?
Thanks,

marks
04-02-2011, 15:14
it seems as OcO hasn't run the tests in a while.

As you can see in the installation folder: /usr/local/oco/
there is folder bin/ with the programs that run to keep the status of the server up to date. These process update the value of the OcO result.
Some of them might be failing, quite possibly because of the IP loadbalancing not pinging from your server.

I'd firstly make sure that the IP load balancing does work.

After, update us.

cioby23
04-02-2011, 12:09
I have installed OCO service in a Load Balancing infrastructure with 2 servers. I have allowed port 79 in firewall. Initially the probe returns ok but after some tries it gives the message OCO Probe: 398 - Result too old. Also the IP of the load balancing (assigned to eth0:0) is not pingable.
I have restarted the service several times and put the service in KO and then OK mode again.
Some can provide some advice ?
Kind regards,