Problems with SIM 5.0 SP4

[Goonboon]

First off let me give you the specs of what version SIM is and what it is installed on.

-SIM Info-
SIM Version: 5.0 SP4 - Windows
Build Version: C.05.00.01.01
Build Date: 2006-02-01 12:06

-Server Info SIM is installed on-
Server Model: Proliant DL385R G1
Windows Version: 2003 Advanced Server
Memory: 4 gig
Processor: Dual AMD Opteron 250 2.41GHZ Dual Core
Database: Local MSDE

Ok here's the problems I am having with SIM.

First off I see SIM having a lot of potential, but from what I am finding out (after working with many different version) is that it is very buggy.

Now I am no newbie at CIM 7 or SIM 5. I have been an admin with these systems for years and from what I am seeing with this latest build is that it's probably the buggiest I have seen.

System Insight Manager GUI Problems:
Ok now I have been working on this latest build: going through and fixing autodiscovery problems, resetting agents, upgrading agents on over 550 devices. This has taken me weeks to get SIM to actually detect things correctly with associations and things looking all nice and pretty :) Unfortunetly I was just wrapping up the cleanup when the whole system decided to unassociate everything I worked on. All servers are now showing up with their MP's unknown. I lost the cluster associations, iLO, and blade associations with their parents.

This is where I'm getting frustrated... It has been a constant battle with HP SIM 5.0 (any build) with something either the database fills up and crashes or SIM deleting systems without being checked for deletion, and now the system unassociating everything.

The other bugs I am seeing can be read below. Any solution as to the unassociation thing that just happened out of the blue would be helpful. Don't really want to lose weeks worth of work.

We deal with many different version of the agents on our servers but most of them are over 7.0. Updating agents on our production servers is not the easiest thing to do when multi million dollar transactions are relying on these servers.


Autodiscovery problems:
Basically Autodiscovery is good for just getting a good view of your subnets and what servers are on it without having to go through and manually entering them all in. There is usually A LOT of cleanup work that still has to be done.

-Clusters
We use Microsoft clustering for our clusters here. From what SIM seems to like to do on 50% or so of the clusters is list one side of a node as "Cluster" in the system type instead of listing it as a server. It also likes to list the cluster node itself as a server. Of course the associations are all messed up when this happens. We usually have to remove the whole cluster from insight manager, add in the cluster node first, lock it down by editing the system properties and forcing a lock, then add in the clustered nodes and thier iLo's. If we don't lock the cluster nodes down first we will have problems adding in one of the nodes which will override the cluster node. In fact we still have clusters that are just not cooperating with SIM and have just decided to leave them out.

-ILO's
Sometimes they are listed as servers, instead of management processors. 50% of the discovery also does not have them associated to their correct parent server. Some will absolutely will not associate with thier parent server.

Thanks,

Chris

[Rob Buxton]

Well I'll say I'm not seeing as many issues.

We only have around 150 servers to manage.
I just use the Automatic Discovery and ensure I include all of the servers, ILO's and some switches in those discovery ranges.
I've learnt from experience that servers with dual IP's cause issues so I use the exclusion lists to limit that.

ILO's need the returned data to be set to medium as a minimum, but once done they seem to associate okay.
If you don't need WBEM you may want to look at disabling it as that still causes some problems.
We don't have any MS Clusters.

We do have VMWare Hosts and their guests and they all show up okay and VMM adds improved managability. And last night we did our first P2V all niccely initiated and monitored via HPSIM.
And finally with our new EVA8000 and SMI-S we're getting quite a bit of storage visibility.
All this and a database of historical disk usage that I can capacity plan from and write personalised interrogation queries to access.
And it's free.......

[Goonboon]

Thanks for the reply Rob! Yeah it takes some playing around to get the servers to be detected right.

The biggest problem I see right now is that the "Managment Processor Unassociation" bug that cropped up in SP3 was suppose to be fixed with SP4 and now it seems to have come back for some of us.

I basically was spending weeks making sure associations were correctly configured and then it decides to unassociate everything.

Just very very frustrating... :(

Link to original problem in SP3:
http://forums1.itrc.hp.com/service/forums/bizsupport/questionanswer.do?threadId=995261

As you can see at the bottom of that forum post there is a reply from someone who is still having the problem in SP4...

[Matthew J Warrick]

I had some initial issues with clusters however I discovered that the agents were disabled in Control Panel. That seemed to fix the ID problems. Then I had problems with getting Cluster Monitor to launch, but it randomly started working so that's all good now. I haven't had any ilo problems with SP4 unless there was a firewall issue or the SNMP settings were misconfigured.

The only real issue I have left is multihomed systems, RAS servers for instance, using the wrong hostnames and generally being fussy.

My current customer uses Win2k3SP1, SIM5.4, and 7.40B mgmt agents.

SIM 5.4 seems to be _much_ better than the other versions I have had to implement over the years. Nothing's perfect though.