Appendix e, Installing multiple agents on same nonstop – HP NonStop G-Series User Manual
Page 98
Appendix E: Installing Multiple Agents on Same NonStop 98
Appendix E: Installing Multiple Agents on Same
NonStop
The Configuration Name parameter in the Host Installation Client specifies the name of the configuration where
the OVNM Agent is installed. This is useful when you want to install multiple OVNM Agent environments on the
same HP NonStop system. However, this can cause problems with Enterprise Management systems. They will not
be able to distinguish the various systems on the HP NonStop, so all commands sent from those systems will be
delivered to a single HP NonStop environment.
You cannot have duplicate environment definitions where the IP and Ports all match. In other words, you cannot
use IP 1.1.1.1 and ports 5020/5030 multiple times. The configuration name cannot be duplicated either. So if
you want to use the name “production” on multiple nodes, you will have to include the node name (or some other
descriptor) to make it unique (For example, Chicago-production, Vegas-production, Miami-production, etc.). The
HIC Configuration Name does not support any wild card entry.
There are other items that need to be addressed as well. For example, you cannot use the same process prefix if
you want both of those environments running at the same time. Be aware that if you do decide to use the same
process prefix for two different environments on the same node, the STOPOVNM job will stop the environment
using that process prefix, no matter which subvolume you are in when you RUN STOPOVNM. You cannot use
the same OVNM subvolume and should probably avoid using the same Work Subvolume for multiple
environments.
Considerations when using multiple environments on the same HP NonStop Server:
a.
Configuration name cannot be duplicated. One method to make them unique is to include the actual
HP NonStop Node Name in the configuration name. For example – Chicago-Production.
b.
When defining the communication path, you are specifying an IP (or DNS name) and a port for the
monitoring side (Config Client port) and a port for the Enterprise Management/Console side of the
application (OVNM port). You cannot duplicate any combination of IP (or DNS name) and a port. So if
you have used IP 2.3.4.5 with port 5020, you cannot specify that IP/port combination on any other
configuration. In other words, each environment is required to have its own UNIQUE paths to
everything it communicates with.
c.
The Enterprise Management system could have issues trying to decide which environment sent which
message. If the only information it has to work with is the HP NonStop Node Name, this will not be
unique enough to determine which environment sent it.
d.
You will not be able to use the default value for the process name prefix (OV) for more than one
environment, unless you never expect to have both of those environments up at the same time. If you do
use the same process prefix for multiple environments, be aware that running the STOPOVNM job from
any subvolume will stop the environment using that process prefix, regardless of what subvolume was
used when the STRTOVNM job was executed. So we highly recommend using unique process name
prefixes for each environment on the same HP NonStop Node.
e.
You will have to review the EMS collectors that are being consumed by the various environments and
decide which environments you want to consume which events. If you have multiple environments that
LOG TO the same EMS collector and they both consume events from that EMS collector, then every
single message will be processed multiple times – once for each environment.
•
The first step to resolve this issue is to create EMS collectors for each environment. This will then be
used as the “Log To Collector” in the Host Installation Client.
•
You need to review all the other EMS collectors that are consumed to determine if you want
multiple environments consuming those events or not. And there are two parts to this issue. There
are two programs that consume events - $xxOCC and $xxEMN. $xxOCC consumes them for the
monitoring portion of the application. If you are not going to be setting EMS thresholds against the
events for that collector, you probably do not need to consume those events. These distributors are
defined in the OCC tab in the Host Installation Client. $xxEMN is the process that sends events to
the Enterprise Management system and Console. If there are multiple $xxEMN processes
consuming the same events, then duplicate messages will be sent to the Enterprise Management
System and Console.
•
If multiple environments have the same threshold defined, then multiple environments will generate
the alert. For example, if they both have CPU PERCENT BUSY > 50% and the CPU is busy; they
will both generate an alert. If you have a recovery defined for that threshold on both environments,
then both environments will initiate the recovery. The same is true for emails as well.