Long term unattended testing considerations
The CX23-R/eDAQXR system is designed to support fully unattended testing in remote locations. The primary functional requirement to support this is the following. Whenever a test run is interrupted for any of the following reasons, the CX23-R/eDAQXR starts a new test run when it reboots.
- Power supply is lost
- Power is switched off (via the front panel push button switch, or a switch connected to the remote power control wires)
- CX23-R/eDAQXR resets on error
- User initiates a CX23-R/eDAQXR reboot using the web browser interface
NOTE
Before starting a test, make sure there is adequate storage available on the CPU for the test, whether the CPU is used as a master or slave node. Less than 2 MB of available storage will not allow a test to start.
Power control switches: In releases prior to v2.0.0, the system could apparently very rarely “lose track” of the On/Off state that is changed each time the momentary contact front panel push button switch is depressed. One instance of this was reported by a customer and it was also seen once in QA testing. Because on this, the option to completely disable the front panel switch and also the remote power switch was added. See Hardware > Controlling a module (Reboot, Flash LED, Update firmware, calibration certificate, etc.) > Disable/Enable power switches for more information. There has been significant rework of the system code responsible for tracking the state of the power switches and while this may not be an issue anymore, that is not known conclusively.
Issues with MX modules: There are multiple issues that can come into play when MX modules are used with the CX23-R/eDAQXR. These issues typically occur very rarely when only 1 or 2 MX modules are connected. However, as the number of connected MX modules increases and the EX23-R switch is used, the likelihood of these issues showing up increases significantly. Note that these issues come into play only when a test run is started after a power up or reboot.
- MX subscription time stamps out of phase tolerance: In brief, this means that one or more of the MX channels would be out of phase by more than the allowable limit of 10 µs if the test run was started. When this happens, the CX23-R/eDAQXR reboots the MX module(s) and then reboots itself. If this happens, about 2.5 minutes of test data will be lost. Typically when this happens, the MX channels phase skew is less than ½ of a sample period. For most testing applications, this is insignificant and because of that, there is a System preference to not reboot the system if the skew is less than ½ of a sample period.
- MX modules never synchronize on power up or reboot: It always takes a little time for the MX modules to synchronize to the CX23-R/eDAQXR system using PTPv2. Typically, the time required is a fraction of a minute. However, sometimes the MX modules take a very long time to achieve synchronization. If all MX modules are not synchronized within 5 minutes, the CX23-R/eDAQXR reboots all MX modules then reboots itself. If this happens, about 7.5 minutes of test data will be lost.
- MX modules are not discovered on power up or reboot: The CX23-R/eDAQXR discovers MX modules by monitoring the Ethernet connection traffic and waiting for the MX modules to provide an “announcement” message indicating that all required services are available. Typically, this takes no more than a few seconds. However, sometimes the proper announcement messages are not received as expected, and in this case the CX23-R/eDAQXR reboots the MX module. If this happens, about 6 minutes of test data will be lost.
Summary
All of the known issues with long term unattended testing have workarounds that will limit data losses to relatively short periods (i.e., less than 10 minutes) when any of these issues comes into play on a system power up or reboot.