So if we add another RHQ Server to that server mix? If we do that, I'd imagine
we'd want the storage nodes to replicate to all servers that are 'paired',
no?
-----Original Message-----
From: rhq-users-bounces(a)lists.fedorahosted.org
[mailto:rhq-users-bounces@lists.fedorahosted.org] On Behalf Of Jay Shaughnessy
Sent: Tuesday, August 12, 2014 11:02 AM
To: rhq-users(a)lists.fedorahosted.org
Subject: Re: F5 GTM - RHQ Agent configuration issue
It really has nothing to do with Storage Nodes. It has only to do with RHQ Servers and
Agents. The RHQ Agents talk only to RHQ Servers. They contact those servers via the
host:port information configured for the
*servers*, and stored in the failover list. I don't see how
manipulating the failover list can help you, as it already contains all of the possible
RHQ Servers, and already orders them in a fashion such that agent load is evenly
distributed amongst the available RHQ Servers.
When you use --cleanconfig and specify the RHQ-Server name, RHQ-Server Port the agent
*only* uses that server information to make first contact with a server. It then gets the
initial failover list from that first server contact. It then drops that connection and
establishes a new server connection based on the failover list.
Only the RHQ Servers communicate with the storage cluster, as part if the back-end
repository of data. The storage clusters are completely hidden from the agents.
On 8/12/2014 10:29 AM, barry.barnett(a)wellsfargo.com wrote:
Do you think if we add the BCP storage node to the cluster, so that the storage nodes are
all replicated, and then also have BCP use the same backend Oracle DB as production, then
the agent would pick up on the new BCP server in the mix?
-----Original Message-----
From: rhq-users-bounces(a)lists.fedorahosted.org
[mailto:rhq-users-bounces@lists.fedorahosted.org] On Behalf Of John
Mazzitelli
Sent: Tuesday, August 12, 2014 9:21 AM
To: rhq-users(a)lists.fedorahosted.org
Subject: Re: F5 GTM - RHQ Agent configuration issue
I do not believe you can do what you want.
First, you *could* manually change the failover.dat file, but:
a) the agent will re-generate that file from information it gets from the server whenever
it connects to a server or on the hour (the agent periodically asks the server if its
failover list changed - you could turn that off by setting the
rhq.agent.primary-server-switchover-check-interval-msecs preference to 0 (see the comments
for that setting in agent-configuration.xml), but the agent will still regenerate it
whenever it connects to a server either at startup or at failover to another server.
Can't tell you how things will work if you set that to 0. If you add a server to the
HA environment, you'd need to re-connect all your agents to get that list to update,
which is why in the comments for that setting we say you shouldn't set it to 0 unless
you only have a single server in your HA env.
b) that file doesn't have the URL - its just the hostname and ports (secure and
unsecure port). The agent uses JBoss/Remoting for its communications API and it uses
JBoss/Remoting protocols (like servlet or sslservlet for example). But if the hostname and
port is all you need to go over your F5 redirector, it may be all you need. I don't
know. But you would then have to contend with a).
The failover list is generated from the server public endpoint - see the UI page
Administration/Server and look at your servers - see the public endpoint information?
Hostname, secure port, port? That's the information that goes in the agent failover
lists. I don't think you can change those to a common F5 URL because that might cause
servers not to startup properly and/or it might cause DB constraints in the RHQ_SERVER
table. I don't know, I've never tried it, but I would think bad things are going
to happen.
In short, I can't think of a way to do what you want. It may be possible by doing
out-of-the-ordinary things like manually editing failover.dat or setting the public
endpoints to all your servers to be the same and turning off the switchover check
interval, but all of those have possible side effects or might not even be possible at
all. So I can't say how it would work.
I recommend reading the following wiki pages that talk about agent communication,
registration, failover, etc.
https://docs.jboss.org/author/display/RHQ/High+Availability#HighAvaila
bility-FailoverLists
https://docs.jboss.org/author/display/RHQ/Communications+Configuration
https://docs.jboss.org/author/display/RHQ/Agent+Registration
----- Original Message -----
>
> Would I simply change the HA parm in the rhq-server.properties file
> to use the F5 URL as opposed to the prod host names?
>
> -----Original Message-----
> From: Barnett, Barry F
> Sent: Tuesday, August 12, 2014 8:39 AM
> To: 'rhq-users(a)lists.fedorahosted.org'; mazz(a)redhat.com
> Subject: RE: F5 GTM - RHQ Agent configuration issue
>
> I found the failover-list.dat file in the rhq-agent/data directory.
> Can I make the change to this file directly? Do I need to make the
> change on every agent I just added (44 of them) to point to the RHQ
> Servers, or is there a way to push out the additional hostname to the
> failover list of all agents?
>
>
>
> -----Original Message-----
> From: Barnett, Barry F
> Sent: Tuesday, August 12, 2014 8:33 AM
> To: rhq-users(a)lists.fedorahosted.org
> Subject: RE: F5 GTM - RHQ Agent configuration issue
>
> After I install the agent, I configure it by issuing the following on
> Windows:
>
> rhq-agent.bat --cleanconfig
>
> When I enter this command, it will ask me for the agent name, host
> server, RHQ-Server name, RHQ-Server Port, etc.. I enter the name of
> the destination RHQ Server as the F5 URL, and its port under the
> RHQ-Server Port. I thought the Agent would then use that to point to
> wherever the F5 was pointing to at that instance in time. If the F5
> pointed to another RHQ server, then I thought that agent instance would also then
point there.
>
> I'm guessing that when I configured this originally, it did use the
> F5 URL, which was connected to Prod Node A at the time. And then the
> agent was passed the failover list only from Prod, and now uses that to connect.
> Could I manually edit this list to add the BCP server??? If so, how
> do I do this?
>
> -----Original Message-----
> From: rhq-users-bounces(a)lists.fedorahosted.org
> [mailto:rhq-users-bounces@lists.fedorahosted.org] On Behalf Of John
> Mazzitelli
> Sent: Monday, August 11, 2014 3:47 PM
> To: rhq-users(a)lists.fedorahosted.org
> Subject: Re: F5 GTM - RHQ Agent configuration issue
>
>> Why wouldn’t the agent that points to the F5 GTM URL point to
>> wherever the F5 wants to have it communicate with? So you’re saying
>> the BCP RHQ server storage node has to join the Prod storage node
>> cluster for this failover to BCP to work, even with the F5 GTM URL being used?
> Don't confuse the storage node configuration with the agent
> configration. Two different things. The agent's failover list is
> determined when a new server is added to the RHQ HA environment and
> shared to the agent the next time the agent connects to the server or
> when the agent periodically asks for its failover list (which happens every hour by
default).
>
> The storage cluster stuff is handled independently/differently from
> that. I dont know much about the storage node cluster config; someone
> else would have to chime in there about storage nodes.
>
> When the agent needs to connect to a server, it looks for the
> hostname and port of the server to use (the data you see in
> failover.dat) and will use the proper Jboss/Remoting protocol (which
> is typically either servlet or sslservlet). So the URL it tries to
> connect to will be some jboss/remoting URL like
>
"servlet://server-hostname:7080/jboss-remoting-servlet-invoker/ServerInvokerServlet"
>
> I don't know how you are telling the agent to use this F5 GTM URL you
> refer to, but I have a feeling the agent isn't using the URL you
> think it is using. But that's just a guess. Turn on agent debug, and
> look at the debug messages it spews when trying to make connections
> during its failover and see what its trying to connect to.
> _______________________________________________
> rhq-users mailing list
> rhq-users(a)lists.fedorahosted.org
>
https://lists.fedorahosted.org/mailman/listinfo/rhq-users
>
_______________________________________________
rhq-users mailing list
rhq-users(a)lists.fedorahosted.org
https://lists.fedorahosted.org/mailman/listinfo/rhq-users
_______________________________________________
rhq-users mailing list
rhq-users(a)lists.fedorahosted.org
https://lists.fedorahosted.org/mailman/listinfo/rhq-users
_______________________________________________
rhq-users mailing list
rhq-users(a)lists.fedorahosted.org