mpse Monitoring Provisioning Service Ecosystem

[Anon Anon]

So as near as I can tell, mpse is part of the adminconsole.  Apparently not very many people at CA know what it does.  Various documents contain fragments that are irrelevent, wrong, outdated or contradictory.

 

https://wiki.ca.com/display/UIM81/Primary+Hub+Component+Reference

 

"Licensing for UIM Snap. In UIM, indicates it is not a snap deployment? It does something? Must be in a deployment. Only on primary?"

 

http://docs.nimsoft.com/prodhelp/en_US/Monitor/7.5/ReleaseNotes/index.htm?toc.htm?2162829.html

 

7.5 claims it should be deployed on all hubs, but that doesn't seem to be the case. Possibly needed on all admin hubs.  Netstat shows it makes connections to the database, so that wouldn't even be possible on a remote probe.

 

https://wiki.ca.com/display/UIMPGA/Probe+Groups#ProbeGroups-SLMGroup

 

The probe groups doc summarizes it as providing config guis for the admin console, but other sources claim thats what ppm does.

 

I've got a case open at support, and so far all they could tell me was that they found a communication diagram which doesn't seem to include a database connection.

 

Guessing on the non-obvious ones:

DE = data_engine

QP = qos_processor

DISCO = discover_server

AC = adminconsole

AE = Alert Enrichment

NAP = ?

WIM = ?

monitoring_services is one of the tomcat webservices used by admin console.

 

Does anyone know what the **** this thing does and how it's supposed to be deployed.  Primary hub only? HA? On every admin console?

 

 

[Anon Anon]

That is an excellent question! One I would like an extensive answer to myself. I'm trying to dechiper the nightmare knows as ppm/mpse/adminconsole as well, which is easier said than done.

 

My end goal at least is to ensure I have numerous admin consoles running, as I don't trust it as a management interface due to circular dependencies. But also to just get the stuff working (easier said than done).

[Anon Anon]

Info from support case.  The guy is doing a good job tracking down some notes in the code and doing some testing.  I think some of his assertions are based on old 7.x documents which contradict 8.x documents in some cases and are ommitted in 8.x documents in others.  8.x actually suggests deploying adminconsole to all hubs and fails to mention the dependency on ppm for on all hubs for admin console.  7.x suggests that both ppm and mpse must be installed on all hubs.  I could have sworn I saw some db connections in netstat, but he didn't find them in his testing.

 

Anyway, props for this persons effort.  They're working it, but I don't think they have gotten access to anyone who knows the answers or the design.  

 

== From support.

 

Some of these answers may be somewhat contradictory to the documentation. Documentation states that mpse must be deployed on each hub where ppm is deployed as a requirement of probe configuration through admin console. In my testing this is not necessarily true and I was not able to find a probe where mpse was specifically required outside of the primary hub.

This doesn't mean the documentation is wrong, it simply means that I wasn't able to find a situation in my testing where it was necessary.

1) Where is it supposed to run
It needs to be deployed wherever there is an admin console. That being said, it is my understanding that multiple admin consoles are not guaranteed support, but it is typically possible to make it work. According to the documentation, it needs to be deployed on any hub alongside ppm.

2) What happens when it stops running
The only side effects that I've seen personally from it not running is that admin console will not allow you to log in. Wasp also makes a connection to mpse, but I'm not certain right off what other wide reaching implications it might have.

3) Should I have only one running in an environment
According to documentation, it is fine to have multiples.

4) Does it belong with primary hub in HA setup to be failed over?
It's fine to have mpse on an HA hub. Of course you run into the issues where wasp points specifically to mpse (and other probes) on the primary, but that's another topic. In an HA failover scenario, however, it doesn't seem to make sense to have mpse deployed to that HA hub unless you are simply adhering to the guidelines of documentation, or have a secondary admin console deployed to that HA hub.

5) Does it need to be on every host with an admin console.
Yes

6) Does it need access to the database?
It doesn't appear as such. It will look for the data_engine when it is running, but after watching SQL Profiler, I can't find that it ever physically touches the database. Additionally, tunnels would complicate database access.

[Anon Anon]

Hi,

I checked yesterday and it does some queries to the db. I checked from the dB, but you can see it from the log, too.

-jon

[Anon Anon]

This is one of the things I find very frustrating.  This senior support engineer, for all his work and patience, is not being given the resources and access to provide good support.  This guy is clearly scanning whatever documentation he can find internally and externally as well as testing the product and poking at the product to try and discover an answer.

 

How much of this his time, his peers time, and our time is being wasted because whomever wrote this thing is gated several layers away from the customer and being flogged to churn out another 3+ unusable, beta-grade releases this year with even more undocumented components and bugs.

[Anon Anon]

For running multiple admin consoles, I got this answer from support (last Aug):

 

"Unfortunately, the KB Article which is referenced in this case is outdated. The method developed to run the service_host remotely was an unofficial procedure which was developed by a support engineer and which was never fully QA'd or considered to be a supported operation.

Unfortunately, in the time since that article was published, there have been changes to the service_host having to do with the Dynamic Threshold and Baseline capabilities which we have integrated into the Admin Console. At the time that article was written service_host did not need to have access to the database, but currently in the newest version of the product, it does require this.

Therefore, at this time, there is no currently supported method for running the Admin Console on a remote hub. The only currently supported and working method is to open the ports on the primary hub firewall and allow access directly to ports 8080/8443 on the primary hub where service_host runs.

We have removed the knowledgebase article referencing this from publication to avoid confusing customers in the future.

We have also reached out to the development team to discuss this further; they will consider this for future development so that it will be once again possible to host the Admin Console on a remote hub, but right now it just can't be done.

[odom]

Ray, I feel your pain.... over the last few years of attempting to work/setup our Nimsoft environment these issues crop up more and more. Not only from simple installations but the probes themselves have issues in where they work sometimes, and sometimes not. Nothing is consistant and at times I feel like paying customers are the QA department.