So you have Discovered, given Access and run Inventory on your HMC and it is not finding the Servers and Virtual Servers that the HMC is connected too! What do you do next?
Well here are a series of suggestions for you to try (this is not an official diagnostics path but might help you out):
How much memory has your HMC got?- I get people trying to use old HMCs and the original ones had just 1 GB of memory.
- This is basically not enough these days - in my humble opinion - unless you have just one machine!
- But how do you find out what you have?
- Opening the HMC to take the RAM out seems so old fashioned!
- So ssh to the HMC (I use putty) log on as hscroot (or equivalent) and run: cat /proc/meminfo
- The top line is memory size.
- If its 1 GB then you may struggle to run the Systems Director agents.
- You might try rebooting your HMC to flush out memory use and free up some memory.
- You can upgrade some HMC with more memory but basically ... good luck!
Have you got a recent HMC software level?- Log on to the HMC and on the Welcome screen - it should tell you the software level.
- Anything below 7.7.2 is "in my opinion" ancient history. OK, you may have Power5 and it has long been stable but running without all the fixes and newer function seems silly to me and you don't get my sympathy when it eventually breaks. You do regular maintenance on your car ... don't you!
- Upgrade to at least 7.2.2 or better yet 7.7.3 (at the time of writing = summer of 2011) but do check your pre-reqs like systems firmware levels.
- Some old HMC software levels had a Systems Director memory leak - we are taking a few years ago. This would later knock the HMC off the network and require a hard reboot in person - just don't go there!
- You really don't want to be running out of date HMC that is like running a new car on bicycle tyres.
- And if there are HMC fixes for your new HMC level - install them too. They are there for a reason.
Now to the Systems Director side - Have you got the latest version?- On the Welcome + Manage screen, the top item is the "IBM Systems Director Server" version.
- While diagnosing a friend's set-up he thought 6.2.1 was good enough. I said "Oh no, my friend, not true."
- I am running 6.2.1.2 - it does not sound like much of a difference but that is two rounds and 9 months of bug fixes.
- That is a major difference. Only a fool runs out of date systems management software.
- Always run the latest Systems Director version once it has been out for 2 weeks (just a sanity check) - I upgrade on the day I notice a new release is available and it has not failed me.
- How to do this? Click "Update IBM® Systems Director" on the same screen - "simple".
- This checks immediately over the Internet, if there is a new version and from there you can update Systems Director - it updates itself.
- Don't expect this to be quick - it is often a large download followed by an update (to set expectations I will just use the Java word). Also some parts are updated when its restarted as it can't be done live.
- Then restart Systems Director as root on its server: smstop; smstart, smstatus -r and go for lunch. On your return, you will see "Active" on the command line screen.
- If you have not got direct Internet access you will have to follow the Redbook/manuals about downloading by hand, transferring the images and updating Systems Director on the command line (I have never needed to do this, so good luck).
I will not insult you with questions like: Can you ping in both directions!Does Systems Director know it is a HMC or not?- With Systems Director take Navigate + All Systems then search for the HMC by name.
- It should have two entries.
- One with a Type of Hardware Management Console and One as the Operating System.
- If the HMC one is missing (typically it might just say Server) then System Director found the endpoint but did not work out it was a HMC.
- It is worth just trying from the top to connect the HMC again.
- Right click on the endpoint then select Delete and confirm the OK.
- Do this for the second endpoint.
- Then go back to find the HMC again with Discover, Access and Inventory.
Has Systems Director used the right protocols for "talking" to the HMC?- Navigate to the HMC endpoint then right click and select Security then Configure Access.
- It should show a few lines of output further down under "your HMC hostname (Remote Service Access Point) - see the below example.
- The Access Type "CIM" line should be green and "OK"
- They may also be Access Type SSH and SLP lines and I have never got a straight answer about if they are vital or necessary or just nice to have.
- Mine say "offline" and "unknown" but my HMC apparently works OK - I guessSystems Director talk to the CIM to do all the bulk of the real work.
- Perhaps, there are specific actions that need these services - I have not found one yet.
What about firewalls and ports?- In the Configure Access Tab (see above on getting there) you should see on the CIM line the port it uses.
- The default is 5989 for CIM to the HMC - good luck if you have to change ports, I read up on it and it sounds a nightmare!
- So, is that port possible between the Systems Director server and the HMC on your network?
- That one is tricky. You can ask your network administrator - if bright they should know the answer.
- I tried a quick and messy approach from the Systems Director server I ran: tn hmc.mydomain.com 5989
- For a valid port telnet (tn) connects and then gets stuck but outputs "Trying... Connected to hmc.mydomain.com. Escape character is '^T'. "
- I think this means the network socket connects but then stops as we are not talking the write protocol.
- For an invalid port, I get just "Trying..."
- I think this means it fails to get the network socket open
- I did say quick and messy :-)
What to do if you have a firewall issue?- Get it fixed, of course - sorry without network connection on this vital port in particular, you are not going to get far.
- But while you are talking to you network administrator, review the other network ports needed.
- For Systems Director endpoints they are documented here http://tinyurl.com/SDports
OK, so now you have:
- Enough memory on the HMC
- A recent HMC level installed
- Latest Systems Director server version and its restarted
- The firewall fixed
Try the HMC connection from the top
- Delete the two HMC endpoints
- Probably overkill but if you are the only user, restart Systems Director to make sure the old HMC details are not cached
- Now Discover
- Gain Access
- Run Inventory
- Then you will find the Servers and Virtual Server that the HMC knows about will appear in the Navigate panels.
Notes:
- The Inventory of the HMC assuming it has a handful of machine connected is surprisingly quick - it talks just a few seconds, don't go assuming (like me) that if failed.
- The found machines (Servers) and LPARs (Virtual Servers) can have odd-ish names. Some times they are named as found on the HMC but other times by the box Serial Number of the machine and the Virtual Servers are the same but with a single digit added to the end which is the LPAR ID number. So you may have to hunt for them.
- Fortunately, Systems Director allows you to rename endpoints to be more meaningful. I would make the HMC and Systems Director names the same or similar you you can drive yourself nuts or make mistakes!
- Most of the above also applies to the SDMC too.
These are a few screen shots of my Systems Directors view of a HMC - it may not be perfect but at least it shows you what a working HMC looks like.
HMC Version 7.7.3 (Not to be confused with 7.3.x) or some guys like to write V7R7.3 but UNIX guys don't do that much typing :-)
HMC Memory Size = 4129152 KB = ~ 4 GB
Systems Director version (current for summer of 2011)
Systems Director Navigate and search for you HMC endpoints
Systems Director Security + Configure Access
I hope this helps, thanks, Nigel Griffiths
Tags: 
power7
aix
powervm
hmc
director
power6
systems