Starting in version 12.55, Performance Center allows you to use Jmeter scripts. To me, this is shocking shift in MicroFocus’s support for open source technologies. While support for Jmeter within Performance Center is in its infancy the impact is significant. Prior to version 12.55, if an organization wants to run large-scale performance tests 10k+ virtual users in performance center they would have 4 options:
- Purchasing permanent licenses <- Very Expensive and not cost effective for infrequent testing. Long term commitment required.
- Purchase term licenses <- Less Expensive, but reasonable if you need many tests for a few months.
- Purchase VUDs <- Very Flexible, expensive.
- Leverage StormRunner Cloud <- Leveraging cloud is not always an option for internally hosted solutions.
With JMeter support in 12.55, and the increasing cost of licensing what motivation do organizations have in spending thousands in licenses? Granted, Jmeter doesn’t support everything VUDs do, and the scripting in two technologies does present challenges: configuration, installation, and training.
To me, this an interesting approach for MF. How much revenue do they generate from VUD purchases and term licensing? What is the thought process behind this change, what is the long-term goal behind this change of heart, and will it impact existing support costs or will a new ‘JMETER’ protocol bundle created.
As a manger I’m excited about this change and the possibilities it creates. I’m also cautiously optimistic to this change and this new attitude of open source support within a largely pay to play platform. I just hope this new feature isn’t met with a new pricing model.
After upgrading to 12.56 I’ve noticed PC Hosts get stuck in an ‘Initializing‘ status. Use the query below if you have a PC Host that is stuck in the ‘Initializing‘ status to get it back in working order.
set [HOST_RUNTIME_STATE] = ‘Idle’
where [HOST_ID] in ‘<Specify Host ID>’
Below are a list of bugs I’ve discovered in PC 12.56. I’ll update this list over the next 30 days as more are identified.
1) Username/Password in Monitor Profiles are not used (Even when credentials exist). In 12.53 you could leave the username and password blank in a Monitor Profile and PC would use the Service account credentials by default to pull counters from a system. In 12.56 the primary service overrides anything you’ve entered and uses the primary service account. This will impact all monitor profiles that use a custom credentials (anything other than the default service account).
2) This isn’t a bug but is noteworthy. If you try to view Analysis files created via 12.56 tests, errors will be thrown if you use Analysis 12.53 loaded. The issue only shows for the Analyzed Results and not the Raw Results.
3) Just like 12.53, My Performance Center requires a large monitor/screen space. Pretty common these days but it can be frustrating when you attempt to load it from VM your RDP’d into and modifying the screen resolution isn’t possible.
4) Pools in 12.56 are also troublesome. The ‘General’ pool that is created by default does not seem to effectively ‘stick’ with some hosts. I recommend creating a new pool if you intend on sharing access across many projects.
After upgrading to PC 12.53, I noticed one project that was not showing transactions passing or failing in the Run Screen window. All transactions showed 0 count.
To fix the issue the Controller Settings need to be modified. (steps are using ALM Explorer 12.53)
- Click the Gear
- Select Performance Center Controller Options
- Update Sample Information at Frequency to 30 seconds
- Update Data Sampling Rate to 30 seconds
- Click OK
If your like me, and have upgraded Performance Center multiple times, you end up with a very messy list of URLs in ALM Explorer. To clean-up the list of URLs in the Address Bar you need to modify Windows Registry.
Windows Registry: HKEY_CURRENT_USER\Software\Mercury Interactive\Quality Center\
Click each of the old keys and delete them as needed.
While using Performance Center 12.53 I encountered an issue where my Test Runs were becoming stuck in ‘Pending Creating Analysis’ whenever the scenario was configured to Collate (vs Collate and Analyze). The Data Processor Queue in PC was stuck. The issue only became worse as numerous runs began to become stuck in this status, effectively creating a log-jam.
To remedy the issue the following queries were executed in Site Admin – PC LAB. This effectively removed the ‘TASKS’ so that the bad runs could be skipped. Be sure to back-up the ALM DBs / ALM Repository before running any update statements in Site Admin.
set DP_PROGRESS_STATUS=’Finished’, DP_PROGRESS_SUBSTATUS=’Terminating’
Where DP_Progress_Status = ‘Pending’
and DP_Operation_Type = ‘Analyzing’
and QC_Project = ‘<ProjectNameHere>’
and QC_RUN_ID = ‘<QCRunIDHere>’
The statement above was used and fixed the issue. I recommend being very specific and testing it with 1 Test Run before running a mass-update on the DP_TASK table.
Change Machine Identity failed: Reason: RunProcessWithLogon: Failed to create process [D:\PCHost\al_agent\bin\alagentservice.exe] with user <My PC User> windows error code .
My team recently was encountering the error above on multiple VMs after a server patching was completed. The maintenance re-enabled UAC which disabled communication from the PC Hosts and ALM which caused the systems to be Non-Operational.
After a little digging, UAC was determined to be the culprit. For Server 2012, update the registry below to disable UAC.
- Browse to HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Policies\System
- In the right pane, double-click EnableLUA
- Change Value Data to 0
- Reboot the Host, Attempt to Re-Configure in the LAB