Hi, Is it possible (perhaps using a script based monitor??) to check if a Windows server can be logged on to? thx, John Bradshaw

Hi John, can you provide us with more details? Do you want to check a network logon? Remote Desktop? Local console logon? Do you want to check this remotely or locally from the monitored server? http://OpsMgr.ru/

Hi Alexey, Sorry for not giving enough details. I want to test if a Windows server is in the hung state. I know that if I cannot log in to a remote server via RDP (ie see the Ctrl+Alt+Del screen), then the server is in a bad way. Using a script, I want to test if the equivalent of pressing Ctrl-Alt-Del and logging in then off, is Ok or not. (I don't want to actually use an RDP gui session, but use a script in SCOM to see if the whole process is OK or not). thx, John

I think this is a not issue to test logon as monitor of a hunged server. If your logon continue longer than one minute it mean that your domain controller is unavailable for this server. If your server is hung then and services is hung, so System Center Management is hung too, right? And you will receive an aler which tells about Agent unavailability.

Hi Bob, The situation we have had twice now with a SQL server is that the server can be pinged, but no one can log in to it (RDP) and applications do not respond. All we see is a light blue screen on the server (NOT bsod). No heartbeat failures occur. Hence why we wish to see if we can actually log on to the server via script and alert if we cannot. Any ideas would be appreciated. Thx, JB

Yes, I undestood, you need to find a solution for the incedent. Ok, can you try to run some agent task from OM console when server is not responding? What will happen?

Hi Bob, I cannot test this situation as the "hung" state of the server cannot be artificially reproduced. I am looking at running a script to telnet to port 3389. it appears that if this is not listening, then RDP is broken and it's a good bet the server has major problems. I'm in the process of seeing how to do this... John Bradshaw

Hi John, If you see a light blue screen on the server (over an RDC connection, right?) then a port query will not help you. Just because if you were able to see a blue screen then an RDC port were definitely open and responsive. The best way to detect this kind of issues is a syntetic transactions with your application (that do not respond). Can you query this app somehow remotely? http://OpsMgr.ru/

I also think you can try to do synthetic. because you say apps dont work either (not just crashed rdp). SO perhaps the SQL can not be queried anymore.. there is a monitor for that (management pack templates section in your authoring view in scom console).Bob Cornelissen - BICTT (My Blog about SCOM) - MVP 2012 and Microsoft Community Contributor 2011 Recipient

Thx Alexey, I wish to do this for ALL Windows servers, so testing for a different Synthetic Transaction won't be feasible. Is there a common Windows Transaction that would apply to all servers?? There must be a way to see if a box is hung. Thx, John Bradshaw

Thx Bob, Not sure what you have in mind here. Can u pls explain a bit further? Ta, JB

Does the "Health Service Heartbeat Failure" Alert generate when a server is "hung". I'm assuming that communication between Health Service and the Health Service Watcher would fail on a hung server (but maybe not).

Unfortunately, no it doesn't. Not all the time. That is part of the problem, it is not reliable. A server may indeed be hung and unresponsive, but the Hearbeat Failure will not always trigger. JB

You can create a 'general' synthetic transaction that would apply to all servers. Something like a WMI query for win32_Processor. But this common (and generic) transaction will inherit all issues you trying to avoid now: You may have a hung server, application isn't responsive and you can't log on to a server over a RDC, but in the same time it sends a heatbeats (Hearbeat Failure will not trigger) and has a good response to your generic transaction. I am understand you do not want to do all that time-consuming and a pretty labour-intensive syntethic transaction design and development. But I believe it's the only way to make a reliable solution.http://OpsMgr.ru/

Good point Alexey. I do undertand that out of memory pool is perhaps the most common cause of Windows servers hanging. Do u know of a way to check for that? http://vmin.wordpress.com/2012/05/20/why-do-windows-servers-hang/ Cheers, John

Available memory is monitored by Core OS MP by default. If you want to check it remotely you may use a WMI query like strComputer = "." Set objWMIService = GetObject("winmgmts:\\" & strComputer & "\root\CIMV2") Set colItems = objWMIService.ExecQuery( _ "SELECT * FROM Win32_PerfRawData_PerfOS_Memory",,48) For Each objItem in colItems Wscript.Echo "-----------------------------------" Wscript.Echo "Win32_PerfRawData_PerfOS_Memory instance" Wscript.Echo "-----------------------------------" Wscript.Echo "AvailableMBytes: " & objItem.AvailableMBytes Next http://OpsMgr.ru/

THe thing I was thinking about first was (because you mentioned SQL and unavailable app) to do a synth transaction on the SQL level. To see if it functions as such. That doesnt cover any and all servers, but is a bit specific to SQL servers. ODBC check. I see you are already looking at common causes of a hang and how to check for it.Bob Cornelissen - BICTT (My Blog about SCOM) - MVP 2012 and Microsoft Community Contributor 2011 Recipient

Thx Bob. Yes. iyt would be more the common cause Iam looking for. We just happened to have 2 SQL boxes hang on us, but I would like a Windows solution. Cheers, John

This could be it Alexey! Maybe I need to look again at that monitor together with your script and check for any servers going at 100% for over 10 mins. Sounds like it might work.....If the O/S can respond when it is already running cpu at 100%??? What do you think?? John

Well... I do not think that server would be able to respond to WMI when it's at a 100% CPU. BUT, it's possible situation when CPU will be a NEAR a 100% at some time. Like a curve 100->99->100->98->100 over a short time like a second or two and this short 'relax' will give a server chance to respond. Anyway, if it's not responding - it's bad, if a server is able to report and reports a 99-100% CPU or memory low - it's bad again. Both cases should alert you. One thing you need to know: a simple remote performance monitor will not alert you if a query failed, it will throw an error instead. You may want to create a script that will run a query and analyze the result itself. http://OpsMgr.ru/

Thx very much Alexey and Bob JB