Tuesday, June 9, 2020

designer/missingid error when importing application definition in ICD 7.6.1 using port forwarding

Background

I encountered a strange error with my IBM Control Desk development environment and wanted to share because I couldn't find any hits on Google.

Problem

The problem I've encountered is the following error when trying to import any application XML into ICD 7.6.1.1:


And the import fails. It does this on all Windows and Linux browsers that I tried (Chrome, IE, Firefox, Edge, new Edge, Chromium, etc.), so it doesn't appear to be a browser problem.

Workaround

I realized that my ICD VM is running on my VMWare host on a NAT network, and I've got port forwarding configured to allow me to access it from systems on my home network. So I figured I would try the same steps using a browser either running on the ICD VM itself or on another VM on the same NAT network. And that works! So the issue is something to do with my configuration of port forwarding or something else in my network AFAIK. More importantly, it's NOT a problem with ICD.

I've used my browser's Developer Tools to analyze what's going on, and I don't see any difference between a failing browser and a working one, so I think there's something going on in ICD. However, I know that this particular configuration isn't one that I'm likely to see at any customer site, especially since I'm using VMWare Workstation to run my VMs. So I'm just documenting this here so there's at least one Google hit on the problem, along with at least one workaround.

Tuesday, April 7, 2020

How to delete virbr0 and virbr0-nic in RHEL7

Background

I have an RHEL 7.6 VM that I clone to have new servers. This VM was installed with the Desktop profile defaults. This installed libvirt, which I normally want in my VMs, but it sometimes gets in the way. Specifically, I needed to configure two DB2 VMs with HADR and TSAM in a test environment, and I found that virbr0 and virbr0-nic don't allow the 'db2haicu' command to successfully complete. So I needed to quickly delete those interfaces.

List of steps

As root:

service libvirtd stop
systemctl disable libvirtd
ip link set virbr0 down
brctl delbr virbr0
ip link delete virbr0-nic

Caveats

I don't know if the above will correctly survive a reboot.

Friday, March 27, 2020

Thursday, March 26, 2020

One solution for VMMSyncData has no values

Background

During an IBM Control Desk (also applies to Maximo) implementation, we found a problem with the VMMSYNC cron task. Specifically, it didn't appear to pull any data from Active Directory. Or if it did, the data wasn't processed.

Symptoms

We had ChangePolling set to 1 in our cron task instance. We enabled DEBUG level logging for the VMMSYNC cron task and saw the following in SystemOut.log:


[3/25/20 19:28:06:374 EDT] 000001ba SystemOut     O 25 Mar 2020 19:28:06:374 [DEBUG] [MAXIMO0c1] [CID-CRON-1618] Synchronizing VMM Users: vmmSubject Subject:

        Principal: defaultWIMFileBasedRealm/myproduser
        Public Credential: com.ibm.ws.security.auth.WSCredentialImpl@16bdee5d
        Private Credential: com.ibm.ws.security.token.SingleSignonTokenImpl@e022bcb
        Private Credential: com.ibm.ws.security.token.AuthenticationTokenImpl@17329f4d
        Private Credential: com.ibm.ws.security.token.AuthorizationTokenImpl@64c05421

[3/25/20 19:28:06:384 EDT] 000001ba SystemOut     O 25 Mar 2020 19:28:06:384 [DEBUG] [MAXIMO0c1] [CID-CRON-1618] Synchronizing VMM Users: VMM search results
<?xml version="1.0" encoding="UTF-8"?>
<sdo:datagraph xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xmlns:sdo="commonj.sdo" xmlns:wim="http://www.ibm.com/websphere/wim">
  <wim:Root>
    <wim:entities xsi:type="wim:PersonAccount">
      <wim:identifier externalName="CN=myproduser,OU=Admin,o=MYCORP.COM"
          repositoryId="MYAD" uniqueId="CN=myproduser,OU=ADMIN,O=MYCORP.COM"
          uniqueName="CN=myproduser,OU=Admin,o=MYCORP.COM"/>
    </wim:entities>
    <wim:controls xsi:type="wim:ChangeResponseControl">
      <wim:checkPoint>
        <wim:repositoryId>MYAD</wim:repositoryId>
        <wim:repositoryCheckPoint>10085177</wim:repositoryCheckPoint>
      </wim:checkPoint>
    </wim:controls>
  </wim:Root>
</sdo:datagraph>
 
[3/25/20 19:28:06:384 EDT] 000001ba SystemOut     O 25 Mar 2020 19:28:06:384 [DEBUG] [MAXIMO0c1] [CID-CRON-1618] syncUser, syncData = <VMMSyncData has no values>


Notice the greenish highlighted text is the entry that should have been seen in the "VMMSyncData", but it says it has no values.

I then looked at the log of a working environment and saw that it was similar, but it actually worked:


[3/25/20 18:57:12:940 EDT] 000001b3 SystemOut     O 25 Mar 2020 18:57:12:940 [DEBUG] [MAXIMO0] [CID-CRON-102377] Synchronizing VMM Users: vmmSubject Subject:
        Principal: defaultWIMFileBasedRealm/myqauser
        Public Credential: com.ibm.ws.security.auth.WSCredentialImpl@ec82289
        Private Credential: com.ibm.ws.security.token.SingleSignonTokenImpl@5b17dd0b
        Private Credential: com.ibm.ws.security.token.AuthenticationTokenImpl@9348468a
        Private Credential: com.ibm.ws.security.token.AuthorizationTokenImpl@447810d4
 
[3/25/20 18:57:12:944 EDT] 000001b3 SystemOut     O 25 Mar 2020 18:57:12:944 [DEBUG] [MAXIMO0] [CID-CRON-102377] Synchronizing VMM Users: VMM search results
<?xml version="1.0" encoding="UTF-8"?>
<sdo:datagraph xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xmlns:sdo="commonj.sdo" xmlns:wim="http://www.ibm.com/websphere/wim">
  <wim:Root>
    <wim:entities xsi:type="wim:PersonAccount">
      <wim:identifier externalName="CN=myqauser,OU=Admin,o=MYCORP.COM" repositoryId="MYAD"
          uniqueId="CN=myqauser,OU=ADMIN,O=MYCORP.COM" uniqueName="CN=myqauser,OU=Admin,o=MYCORP.COM"/>
      <wim:changeType>modify</wim:changeType>
      <wim:uid>myqauser</wim:uid>
    </wim:entities>
    <wim:controls xsi:type="wim:ChangeResponseControl">
      <wim:checkPoint>
        <wim:repositoryId>MYAD</wim:repositoryId>
        <wim:repositoryCheckPoint>670244</wim:repositoryCheckPoint>
      </wim:checkPoint>
    </wim:controls>
  </wim:Root>
</sdo:datagraph>

[3/25/20 18:57:12:944 EDT] 000001b3 SystemOut     O 25 Mar 2020 18:57:12:944 [DEBUG] [MAXIMO0] [CID-CRON-102377] syncUser, syncData = uid=myqauser
[3/25/20 18:57:12:956 EDT] 000001b3 SystemOut     O 25 Mar 2020 18:57:12:956 [DEBUG] [MAXIMO0] [CID-CRON-102377] insertRecord, columnName = personid
[3/25/20 18:57:12:956 EDT] 000001b3 SystemOut     O 25 Mar 2020 18:57:12:956 [DEBUG] [MAXIMO0] [CID-CRON-102377] value, value = myqauser
[3/25/20 18:57:12:956 EDT] 000001b3 SystemOut     O 25 Mar 2020 18:57:12:956 [DEBUG] [MAXIMO0] [CID-CRON-102377] insertRecord, columnName = type
[3/25/20 18:57:12:957 EDT] 000001b3 SystemOut     O 25 Mar 2020 18:57:12:957 [DEBUG] [MAXIMO0] [CID-CRON-102377] value, value = WORK
[3/25/20 18:57:12:957 EDT] 000001b3 SystemOut     O 25 Mar 2020 18:57:12:957 [DEBUG] [MAXIMO0] [CID-CRON-102377] insertRecord, columnName = isprimary
[3/25/20 18:57:12:957 EDT] 000001b3 SystemOut     O 25 Mar 2020 18:57:12:957 [DEBUG] [MAXIMO0] [CID-CRON-102377] value, value = 1
[3/25/20 18:57:12:957 EDT] 000001b3 SystemOut     O 25 Mar 2020 18:57:12:957 [DEBUG] [MAXIMO0] [CID-CRON-102377] insertRecord, columnName = emailaddress
[3/25/20 18:57:12:958 EDT] 000001b3 SystemOut     O 25 Mar 2020 18:57:12:957 [DEBUG] [MAXIMO0] [CID-CRON-102377] value, value = null


Notice the greenish highlighted text again, where it shows that the bind user was modified, and then it pulls in all of the attributes and values for that entity.

I found this link online and tried what it suggested, but this is outdated (I'm on ICD 7.6.1.1):


That "changeType" attribute is absolutely NOT required any more - it's built into the VMMSYNC code.

The Solution

It turns out that the bind DN (user ID) that we were using apparently didn't have some permission or configuration in place. We actually tried with three different users before we found one that had the appropriate access to actually pull the data. 

What's really strange is that we could successfully use any of the three users with Apache Directory Studio to connect to Active Directory to see any and all data. But for some reason, only one of the users actually worked through VMMSYNC.

I will update if I find out from the AD team what the specific setting is, but I wanted to document this here for posterity.

Monday, March 16, 2020

Integrating IBM Control Desk or Maximo with QRadar for monitoring logins

Background

You may need to monitor ICD/Maximo logins with QRadar, and that's what this post will detail, as there are a few steps and a lack of documentation around it. In this case, I'm using ICD 7.6.1.1 and QRadar Community Edition 7.3.3 (downloaded on 3/13/2020).

All of the steps here apply to both ICD and Maximo, since there is nothing ICD-specific about them, so I'm going to refer to the product as just ICD in the remainder of this post.

Introduction

ICD doesn't log logins by default, and even once you configure it to track logins, it logs them to a database table rather than SystemOut.log. 

Tracking logins

To enable login tracking, you need to go into the Security->Security Groups application in ICD. From there, choose the Security Controls action. In the dialog shown, select "Enable Login Tracking?" and you're done. That's it. To test it, log out and back in again, then use your favorite database tool to look at the contents of the MAXIMO.LOGINTRACKING table. You should see your login.

Configuring QRadar for a JDBC Log Source

This part is a little tricky, but that's why I'm writing this article. You need to create a new Log Source with the following important characteristics:

Type: Universal DSM
Protocol: JDBC
Database name: MAXDB76
Table name: MAXIMO.LOGINTRACKING
Compare field: ATTEMPTDATE

Those are all of the hard properties. The others should be fairly straightforward, such as hostname, port, etc. After that, deploy your changes. Once that's done, log out and back into ICD, and you should get an event that looks similar to this:


Now you can set up the normal parsing, extracting, etc. to have QRadar properly categorize your events.

Tuesday, March 3, 2020

Finding unique element types in XML with PowerShell

I recently needed to deal with a large (165MB) XML file, so the first question I had is: How many unique element types are actually in the file? I found a bunch of truly painful-sounding answers online, mainly involving the distinct-values() XPATH 2.0 function. If you've ever dealt with XPATH, you know that just setting up the toolchain environment to use it is pretty labor intensive. However, I then started looking for PowerShell cmdlets that work with XML and I found Select-Xml:

https://docs.microsoft.com/en-us/powershell/module/microsoft.powershell.utility/select-xml?view=powershell-7

It looks nice enough, but it turns out that it has a few features that are really nice. Specifically, once you read in an XML document, your variable will have a Node member, and that member contains one member for each type of element in your document! So it automatically shows you the list of distinct elements. You can then maneuver through the members to find the number of each type of element ($myvar.Node.myelementType.count) or to view one of the instances ($myvar.Node.myelementType[0]), etc.

In my case, I found that my 165MB file (produced by the ITNM DLA) only contained about 17 unique element types, and that there were only about 2500 network devices, which I could then easily loop through to get the information I needed. Chained with ConvertTo-Csv, I was able to massage the data exactly as needed.

The moral of the story is: PowerShell is amazing right out of the box. I didn't have to add any new packages or anything, and it gave me precisely what I needed.

Thursday, February 27, 2020

JRExecAction() function behavior in Netcool Impact

Just a short post today to set the record straight on the use of the JRExecAction() function in Netcool Impact policies.

The product documentation states that the JRExecAction() function returns nothing, but sets the variables ExecError and ExecOutput corresponding to stderr and stdout from the script/command that you run. However, the function itself returns the data written to stdout in addition to setting ExecOutput and/or ExecError. So if you try the following, you'll see that the same output is logged twice:


results = JRExecAction("/bin/ls", "/", false, 1000);
log(0,"Frank results = " + results);
log(0,"ExecOutput results = " + ExecOutput); 


Incidentally, the first use above is really the more common use in practice. You will find very few implementations where the ExecOutput variable is referenced.