Hi all,
hanging around a Time gap error that I initially attributed to a lack of clock sync between hosts (that was effective, indeed, but not sufficiant). I need to adjust this delay to an higher value...
if someone knows...
thanks.
We actually setup ntp service so all our host share the same clock.
We will survey the behaviour to check if it is enough. Our experience showed that effectively plesiochronous clock setting in a networked system was not stable enough to guarantee avoidance of divergence.
As our XML-RPC goes through a load balancer, and could be routed to hosts with an unkown immediate load, there is a risk that a XML-RPC query would consume more than 15 secs. to be handled at the receiver location. In this case we break the service.
As in our situation, we use a 40 mnetworked Moodle array within a coherent nationwide system, so there are "intimous" XML-RPC exchanges between platforms to federate contents. So we need XML-RPC calls to be performed, including with some delay. The resolution of the call is more important than the delay it takes, as long as it can logically be concluded.
Cheers.
Thanks Nigel. I agree with you that 15 seconds gap for an XML-RPC chain processed on the same servers (literally localhost, including using local network LB gateway seems being so long !!).
I asked my system adminsitrator to rip out Xorg Xwindows task and to lower the running level not to be bothered by those unuseful things... I guess it will help hosts to raise power. We are just on an evaluation system running with low range equipment VMWares so we may anticipate high load effects on future production servers.
Cheers.