> I had posted the full log in the initial message. There is no exception
> stack.
My bad, I thought that was a grep for job/thread ID, I now realize it wasn't.
> If there was an exception stack I probably wouldn't ask stupid questions
> here. :)
So you do realize why I was so keen on the exception stack as it would have
lead us straight to the culprit.
> Thank you! So, should I change all offerings to "static" ones?
Without the exception stack it would be more of a calculated guess /
hit-n-trial. I would may be just test this scenario by temporarily update the
RAM/CPU of the offering (I know custom offerings should have them as NULL).
Since you mentioned that you didn't have dynamic SOs, I strongly suspect the
culprit would be one of so.getRamSize() or so.getCpu(), see below:
reservedMemory += ((so.getRamSize() * 1024L * 1024L) /
ramOvercommitRatio) * clusterRamOvercommitRatio;
reservedCpu += (so.getCpu() * so.getSpeed() / cpuOvercommitRatio) *
clusterCpuOvercommitRatio;
If the above is indeed true, then this would be similar to -
https://issues.apache.org/jira/browse/CLOUDSTACK-8328.
Regards,
Somesh
-----Original Message-----
From: Vladimir Melnik [mailto:[email protected]]
Sent: Sunday, August 23, 2015 2:45 AM
To: [email protected]
Subject: Re: The agent doesn't reconnect if there are stopped VMs
Hello,
On Fri, Aug 21, 2015 at 04:41:42PM +0000, Somesh Naidu wrote:
> > If you mean offerings with "custom" parameters - yes, I use them
> There is a possibility that this might be affecting this.
Thank you! So, should I change all offerings to "static" ones?
> > I don't have more information in the log-file
> Can you open the mgmt server log from "2015-08-18", search for "2015-08-18
> 06:24:46,544", the next few lines should contain the java exception stack
> which should point us to the method-file-line_number that will help us
> understand what is causing the NPE.
I had posted the full log in the initial message. There is no exception stack.
There are 3 lines:
1: 2015-08-18 06:24:46,544 ERROR [c.c.a.m.AgentManagerImpl]
(AgentConnectTaskPool-213:ctx-76903ef6) Monitor ComputeCapacityListener says
there is an error in the connect process for 27 due to null
2: java.lang.NullPointerException
3: 2015-08-18 06:24:46,544 INFO [c.c.a.m.AgentManagerImpl]
(AgentConnectTaskPool-213:ctx-76903ef6) Host 27 is disconnecting with event
AgentDisconnected
The 2nd line doesn't contain an exception stack, all it contains is
"java.lang.NullPointerException".
If there was an exception stack I probably wouldn't ask stupid questions here.
:)
--
V.Melnik
P.S. I've opened an issue: https://issues.apache.org/jira/browse/CLOUDSTACK-8747
