What the test covers: The 1GigEthernet test is run on all wired Ethernet connections with a maximum connection speed of 1 gigabit/sec. Connection speed is determined by parsing the "Speed" line in the output of ethtool.

What the test does: This test adds link speed detection to the existing network test. In addition to passing all the existing network test items, this test must detect a throughput of 1Gb/s (with a margin for overhead) in order to pass. Please see Section A.1.1.27, “network” for information on the rest of the test functionality.

What the test covers: The 10GigEthernet test is run on all wired Ethernet connections with a maximum connection speed of 10 gigabits/sec. Connection speed is determined by parsing the "Speed" line in the output of ethtool.

What the test does: This test adds link speed detection to the existing network test. In addition to passing all the existing network test items, this test must detect a throughput of 10Gb/s (with a margin for overhead) in order to pass. Please see Section A.1.1.27, “network” for information on the rest of the test functionality.

What the test covers: The 20GigEthernet test is run on all wired Ethernet connections with a maximum connection speed of 20 gigabits/sec. Connection speed is determined by parsing the "Speed" line in the output of ethtool.

What the test does: This test adds link speed detection to the existing network test. In addition to passing all the existing network test items, this test must detect a throughput of 20Gb/s (with a margin for overhead) in order to pass. Please see Section A.1.1.27, “network” for information on the rest of the test functionality.

What the test covers: The 25GigEthernet test is run on all wired Ethernet connections with a maximum connection speed of 25 gigabits/sec. Connection speed is determined by parsing the "Speed" line in the output of ethtool.

What the test does: This test adds link speed detection to the existing network test. In addition to passing all the existing network test items, this test must detect a throughput of 25Gb/s (with a margin for overhead) in order to pass. Please see Section A.1.1.27, “network” for information on the rest of the test functionality.

What the test covers: The 40GigEthernet test is run on all wired Ethernet connections with a maximum connection speed of 40 gigabits/sec. Connection speed is determined by parsing the "Speed" line in the output of ethtool.

What the test does: This test adds link speed detection to the existing network test. In addition to passing all the existing network test items, this test must detect a throughput of 40Gb/s (with a margin for overhead) in order to pass. Please see Section A.1.1.27, “network” for information on the rest of the test functionality.

What the test covers: The 50GigEthernet test is run on all wired Ethernet connections with a maximum connection speed of 50 gigabits/sec. Connection speed is determined by parsing the "Speed" line in the output of ethtool.

What the test does: This test adds link speed detection to the existing network test. In addition to passing all the existing network test items, this test must detect a throughput of 50Gb/s (with a margin for overhead) in order to pass. Please see Section A.1.1.27, “network” for information on the rest of the test functionality.

What the test covers: The 100GigEthernet test is run on all wired Ethernet connections with a maximum connection speed of 100 gigabits/sec. Connection speed is determined by parsing the "Speed" line in the output of ethtool.

What the test does: This test adds link speed detection to the existing network test. In addition to passing all the existing network test items, this test must detect a throughput of 100Gb/s (with a margin for overhead) in order to pass. Please see Section A.1.1.27, “network” for information on the rest of the test functionality.

What the test covers: Removable sound cards and integrated sound devices are tested with the audio test. The test is scheduled when the hardware detection routines find the following strings in the udev database:

E: SUBSYSTEM=sound
E: SOUND_INITIALIZED=1

You can see these strings and the strings that trigger the scheduling of the other tests in this guide in the output of the command udevadm info --export-db.

What the test does: The test plays a prerecorded sound (guitar chords or a recorded voice) while simultaneously recording it to a file, then it plays back the recording and asks if you could hear the sound.

Preparing for the test: Before you begin your test run, you should ensure that the audio test is scheduled and that the system can play and record sound. Contact your support contact at Red Hat for further assistance if the test does not appear on a system with installed audio devices. If the test is correctly scheduled, continue on to learn how to manually test the playback and record functions of your sound device.

With built-in speakers present or speakers/headphones plugged into the headphone/line-out jack, playback can be confirmed before testing in these ways:

If no sound can be heard, ensure that the speakers are plugged in to the correct port. You can use any line-out or headphone jack (we have no requirement for which port you must use). Make sure sound is not muted and try adjusting the volume on the speakers and in the operating system itself.

If the audio device has record capabilities, these should also be tested before attempting to run the test. Plug a microphone into one of the Line-in or Mic jacks on the system, or you can use the built-in microphone if you are testing a laptop. Again, we don’t require you to use a specific input jack; as long as one works, the test will pass.

Contact your support person if you are unable to either hear sound or see the input level display move, as this will lead to a failure of the audio test. If you are able to successfully play sounds and see movement on the input level display when making sounds near the microphone, continue to the next section to learn how to run the test.

Executing the test: The audio test is interactive. Before you execute a test run that includes an audio test, connect the microphone you used for your manual test and place it in front of the speakers, or ensure that the built-in microphone is free of obstructions. Alternatively, you can connect the line-out jack directly to the mic/line-in jack with a patch cable if you are testing in a noisy environment. Check the box next to the test name to indicate it is among the tests to run. Click the button Run Selected to continue. The interactive steps are as follows:

Run time: The audio test takes less than 1 minute for simultaneous playback and record, then the playback of the recorded sound. The required info test will add about a minute to the overall run time.

What the test covers: The battery test is only valid for systems that can be powered by a built-in battery use an AC adapter. It does not test external batteries like those found in a UPS, additional internal batteries like the BIOS battery or battery-backed cache, or any other kind of battery that is not providing primary, internal power to the system. The test is scheduled when the hardware detection routines find the following string in the udev database:

POWER_SUPPLY_TYPE=Battery

What the test does: The test does all its work based on the status of the AC adapter. Testing begins with the AC adapter attached to the system. The test scripts verify the status of the AC adapter and that the battery is present. Then the tester is asked to unplug the adapter, which will cause the battery to begin discharging. The test scripts verify this.

Preparing for the test: The battery test requires that the system be connected via an AC adapter when the test is launched. Ensure that it is connected before proceeding.

Executing the test: The battery test is interactive. Check the box next to the test name to indicate it is among the tests to run. Click the button Run Selected to continue. When the test begins, it will display the current status of the battery (capacity and charging status) and ask for the AC adapter to be unplugged until the battery discharges for 10 mWh. The test will automatically end at that point and the tester should plug the AC adapter back in.

Run time: The time of the battery test varies depending on the discharge and recharge speeds of the battery. It takes about 3 minutes on a 2012-era laptop that emphasizes portability and long battery life over screen size and computing power. Because this test is run on laptops, a suspend test must accompany the required info test for each run. The suspend test will add approximately 6 minutes to each test run, and info will add another minute.

What the test covers: All supported optical drives, regardless of formats and features, use the same test methodology, so we are covering all of them in a single section. There are three certification tests for optical media:

Any other disc formats or features like dual-layer (DL) discs, -RAM discs or HD-DVD discs are not tested by the rhcert suite, and can be ignored. The rhcert application determines which of the optical drive tests to schedule, if any, and what type of media to request based on udev information. Here’s an example of the udev database on a desktop computer, showing the supported media of the system’s CD-RW, DVD+/-RW, BD-RE drive:

E: ID_CDROM=1
E: ID_CDROM_CD=1
E: ID_CDROM_CD_R=1
E: ID_CDROM_CD_RW=1
E: ID_CDROM_DVD=1
E: ID_CDROM_DVD_R=1
E: ID_CDROM_DVD_RW=1
E: ID_CDROM_DVD_RAM=1
E: ID_CDROM_DVD_PLUS_R=1
E: ID_CDROM_DVD_PLUS_RW=1
E: ID_CDROM_DVD_PLUS_R_DL=1
E: ID_CDROM_BD=1
E: ID_CDROM_BD_R=1
E: ID_CDROM_BD_RE=1

The scripts look for ID_CDROM=1 before scheduling any of the three optical media tests. If it finds this value, it analyzes the properties to determine which of the three tests to schedule. You can see the drive’s ID_CDROM properties in the udev output above. These tell the rhcert application that the drive is capable of writing to many different disc formats including CD, DVD and Blu-Ray (BD). From that information we know that the bluray, cdrom and dvd tests will be scheduled, and the test harness decides which feature of the format to test. The following tables explain how the rhcert application makes that determination:

The test suite always attempts to schedule the most advanced media tests first in accordance with the rules in the Policy Guide, which requires testing read, write and erase functionality when all are present. Discs that support rewrite functions include:

Only formats supported by the drive are scheduled for testing. If your drive(s) support DVD-RW and DVD+RW, you can use either format of disc during the test. You do not have to test both.

If the drive is not capable of rewrite operations but it does have write-once capabilities for a disc format, the test suite schedules a write-once media test. Discs that support write-once functionality include:

Only formats supported by the drive are scheduled for testing. If your drive(s) support DVD-R and DVD+R, you can use either format of disc during the test. You do not have to test both.

If the drive is not capable of rewrite or write-once operations but it does have read-only support for a disc format, the test suite schedules a read-only media test. Discs that are read-only include:

Only formats supported by the drive are scheduled for testing.

Using the udev data from our example laptop BD/DVD/CD drive from above, we can use this list of discs and tests to determine what types of media are needed. The drive supports all types of Blu-Ray media and since rewritable discs take precedence over write-once or read-only, a BD-RE disc will be needed for the bluray test.

The policy guide was updated at the launch of Red Hat Enterprise Linux 6.3 to reduce the number of optical drive tests that must be performed. Now each controller will only need one test instead of multiple tests of different disc formats. For the example drive shown above, you would run a Blu-Ray rewritable disc test and nothing else. The other tests (CDROM and DVD) are still planned by the rhcert tool, but you do not have to run them. How do you know what drive and which disc type to test? Here is a handy table that explains how it works:

Fill out the table with all the drives you have available to you on your controller. Place an "X" in the column that corresponds with the disc format that each drive supports. When you have finished, choose the drive that has an "X" in the column furthest to the left for your certification testing and be prepared to test that kind of media in the drive. If two or more drives have an "X" in the same leftmost column, you can use either drive for your tests.

Here’s an example.

For the series of drives in the example chart above, you would choose to do your test with Drive 1, and you would only need to run the bluray test with a BD-RE disc. This is because Drive 1 is the drive with an "X" in the furthest column to the left, and that column corresponds with BD-RE media. No other testing would be required.

What the test does: For read-only drives, it reads data from the disc and copies it to the hard drive. The original data on the disc is then compared to the copy on the hard drive. If all file checksums match, the test passes. Writable media adds a write procedure to the test. A blank writable disc is inserted in the system and data is written to it from the hard drive. The data on the disc is then compared to the data on the hard drive. If the file checksums match, the test passes. Rewritable media adds a disc blank to the procedure, followed by a write of data from the hard drive and a comparison of the written data to the original. If the blank is successful and the checksums of the newly written files on the disc match those on the hard drive, the test passes. The test also includes disc ejects between each phase (blank, write, compare). The tester will need to insert the disc back into the drive if the drive is not capable of closing the tray by itself, or if it is a slot loading drive.

Executing the test: The bluray test is interactive. Install the proper drive as determined by the table you created. Check the box next to the test name to indicate it is among the tests to run. Click the button Run Selected to continue. Follow the directions on screen and choose the proper disc format when prompted (the one corresponding with the leftmost column in the table that has an "X" in it), then insert the correct disc when asked. As the test enters the various phases (blank, write, compare, where applicable), the on-screen display will explain what is happening.

Run time: The run time for all optical drive testing is dependent on the speed of the media and drive. For a 4x DVD-RW disc, the DVD test takes about 10 minutes to write and verify ~1.7GB of data.

CD drives of all kinds are tested using the same procedures as Blu-Ray drives. Please see Section A.1.1.10, “bluray” for more information.

What the test covers: The core test examines the system’s CPUs and ensures that they are capable of functioning properly under load.

What the test does: The core test is actually composed of two separate routines. The first test is designed to detect clock jitter. Jitter is a condition that occurs when the system clocks are out of sync with each other. The system clocks are not the same as the CPU clock speed, which is just another way to refer to the speed at which the CPUs are operating. The jitter test uses the getimeofday() function to obtain the time as observed by each logical CPU and then analyzes the returned values. If all the CPU clocks are within .2 nanoseconds of each other, the test passes. The tolerances for the jitter test are very tight. In order to get good results it’s important that the rhcert tests are the only loads running on a system at the time the test is executed. Any other compute loads that are present could interfere with the timing and cause the test to fail. The jitter test also checks to see which clock source the kernel is using. It will print a warning in the logs if an Intel processor is not using TSC, but this will not affect the PASS/FAIL status of the test.

The second routine run in the core test is a CPU load test. It’s the test provided by the required stress package. The stress program, which is available for use outside the rhcert suite if you are looking for a way to stress test a system, launches several simultaneous activities on the system and then monitors for any failures. Specifically it instructs each logical CPU to calculate square roots, it puts the system under memory pressure by using malloc() and free() routines to reserve and free memory respectively, and it forces writes to disk by calling sync(). These activities continue for 10 minutes, and if no failures occur within that time period, the test passes. Please see the stress manpage if you are interested in using it outside of hardware certification testing.

Preparing for the test: The only preparation for the core test is to install a CPU that meets the requirements that are stated in the Policy Guide.

Executing the test: The core test is non-interactive. Check the checkbox next to the test and click the Run Selected button to perform the test.

Run time, bare-metal: The core test itself takes about 12 minutes to run on a bare-metal system. The jitter portion of the test takes a minute or two and the stress portion runs for exactly 10 minutes. The required info test will add about a minute to the overall run time.

Run time, full-virt guest: The fv_core test takes slightly longer than the bare-metal version, about 14 minutes, to run in a KVM guest. The added time is due to guest startup/shutdown activities and the required info test that runs in the guest. The required info test on the bare-metal system will add about a minute to the overall run time.

What the test covers: The cpuscaling test examines a CPU’s ability to increase and decrease its clock speed according to the compute demands placed on it.

What the test does: The test exercises the CPUs at varying frequencies using different scaling governors (the set of instructions that tell the CPU when to change to higher or lower clock speeds and how fast to do so) and measures the difference in the time that it takes to complete a standardized workload. The test is scheduled when the hardware detection routines find the following directories in /sys containing more than one cpu frequency:

/sys/devices/system/cpu/cpuX/cpufreq

The cpuscaling test is planned once per package, rather than being listed once per logical CPU. When the test is run, it will determine topology via /sys/devices/system/cpu/cpuX/topology/physical_package_id, and run the test in parallel for all the logical CPUs in a particular package.

The test procedure for each CPU package is as follows:

The test uses the values found in the sysfs filesystem to determine the maximum and minimum CPU frequencies. You can see these values for any system with this command:

# cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_frequencies

There will always be at least two frequencies displayed here, a maximum and a minimum, but some processors are capable of finer CPU speed control and will show more than two values in the file. Any additional CPU speeds between the max and min are not specifically used during the test, though they may be used as the CPU transitions between max and min frequencies. The test procedure is as follows:

Now the analysis is performed on the three subsections. In steps one through eight we obtain the pi calculation times at maximum and minimum CPU speeds. The difference in the time it takes to calculate pi at the two speeds should be proportional to the difference in CPU speed. For example, if a hypothetical test system had a max frequency of 2GHz and a min of 1GHz and it took the system 30 seconds to run the pi calculation at max speed, we would expect the system to take 60 seconds at min speed to calculate pi. We know that for various reasons perfect results will not be obtained, so we allow for a 10% margin of error (faster or slower than expected) on the results. In our hypothetical example, this means that the minimum speed run could take between 54 and 66 seconds and still be considered a passing test (90% of 60 = 54 and 110% of 60 = 66).

In steps nine through eleven, we test the pi calculation time using the ondemand governor. This confirms that the system can quickly increase the CPU speed to the maximum when work is being done. We take the calculation time obtained in step eleven and compare it to the maximum speed calculation time we obtained back in step five. A passing test has those two values differing by no more than 10%.

In steps twelve through fourteen, we test the pi calculation using the performance governor. This confirms that the system can hold the CPU at maximum frequency at all times. We take the pi calculation time obtained in step 14 and compare it to the maximum speed calculation time we obtained back in step five. Again, a passing test has those two values differing by no more than 10%.

An additional portion of the cpuscaling test runs when an Intel processor with the TurboBoost feature is detected by the presence of the ida CPU flag in /proc/cpuinfo. This test chooses one of the CPUs in each package, omitting CPU0 for housekeeping purposes, and measures the performance using the ondemand governor at maximum speed. It expects a result of at least 5% faster performance than the previous test, when all the cores in the package were being tested in parallel.

Preparing for the test: To prepare for the test, ensure that CPU frequency scaling is enabled in the BIOS and ensure that a CPU is installed that meets the requirements explained in the Policy Guide.

Executing the test: The cpuscaling test is non-interactive. Check the checkbox next to the test and click the Run Selected button to perform the test.

Run time: The cpuscaling test takes about 42 minutes for a 2013-era, single CPU, 6-core/12-thread 3.3GHz Intel-based workstation running Red Hat Enterprise Linux 6.4, AMD64 and Intel 64. Systems with higher core counts and more populated sockets will take longer. The required info test will add about a minute to the overall run time.

DVD drives of all kinds are tested using the same procedures as Blu-Ray drives. Please see Section A.1.1.10, “bluray” for more information.

What the test covers: The Ethernet test only appears when the speed of a network device is not recognized by the test suite. This may be due to an unplugged cable or some other fault is preventing the proper detection of the connection speed. Please exit the test suite, check your connection, and run the test suite again when the device is properly connected. If the problem persists, contact your Red Hat support representative for assistance.

The example below shows a system with two gigabit Ethernet devices, eth0 and eth1. Device eth0 is properly connected, but eth1 is not plugged in.

The output of the ethtool command shows the expected gigabit Ethernet speed of 1000Mb/s for eth0:

# ethtool eth0
Settings for eth0:
	Supported ports: [ TP ]
	Supported link modes:   10baseT/Half 10baseT/Full
	                        100baseT/Half 100baseT/Full
	                        1000baseT/Full
	Supported pause frame use: No
	Supports auto-negotiation: Yes
	Advertised link modes:  10baseT/Half 10baseT/Full
	                        100baseT/Half 100baseT/Full
	                        1000baseT/Full
	Advertised pause frame use: No
	Advertised auto-negotiation: Yes
	Speed: 1000Mb/s
	Duplex: Full
	Port: Twisted Pair
	PHYAD: 2
	Transceiver: internal
	Auto-negotiation: on
	MDI-X: on
	Supports Wake-on: pumbg
	Wake-on: g
	Current message level: 0x00000007 (7)
			       drv probe link
	Link detected: yes

But on eth1 the ethtool command shows an unknown speed, which would cause the Ethernet test to be planned.

# ethtool eth1
Settings for eth1:
	Supported ports: [ TP ]
	Supported link modes:   10baseT/Half 10baseT/Full
	                        100baseT/Half 100baseT/Full
	                        1000baseT/Full
	Supported pause frame use: No
	Supports auto-negotiation: Yes
	Advertised link modes:  10baseT/Half 10baseT/Full
	                        100baseT/Half 100baseT/Full
	                        1000baseT/Full
	Advertised pause frame use: No
	Advertised auto-negotiation: Yes
	Speed: Unknown!
	Duplex: Unknown! (255)
	Port: Twisted Pair
	PHYAD: 1
	Transceiver: internal
	Auto-negotiation: on
	MDI-X: Unknown
	Supports Wake-on: pumbg
	Wake-on: g
	Current message level: 0x00000007 (7)
			       drv probe link
	Link detected: no

What the test covers: The expresscard test looks for devices with both types of ExpressCard interfaces, USB and PCI Express (PCIe), and confirms that the system can communicate through both. ExpressCard slot detection is not as straightforward as detecting other devices in the system. ExpressCard was specifically designed to not require any kind of dedicated bridge device. It’s merely a novel form factor interface that combines PCIe and USB. Because of this, there is no specific "ExpressCard slot" entry that we can see in the output of udev. We decided to schedule the test on systems that contain a battery, USB and PCIe interfaces, as we have seen no devices other than ExpressCard-containing laptops with this combination of hardware.

What the test does: The test first takes a snapshot of all the devices on the USB and PCIe buses using the lsusb and lspci commands. It then asks the tester how many ExpressCard slots are present in the system. The tester is asked to insert a card in one of the slots. The system scans the USB and PCIe buses and compares the results to the original lsusb and lspci output to detect any new devices. If a USB device is detected, the system asks you to remove the card and insert a card with a PCIe interface into the same slot. If a PCIe-based card is detected, the system asks you to remove it and insert a USB-based card into the same slot. If a card is inserted with both interfaces (a docking station card, for example), it fulfills both testing requirements for the slot at once. This procedure is repeated for all slots in the system.

Preparing for the test: You will need ExpressCard cards with USB and PCIe buses. This can be two separate cards or one card with both interfaces. Remove all ExpressCard cards before running the test.

Executing the test: The expresscard test is interactive. Check the box next to the test name to indicate it is among the tests to run. Click the button Run Selected to continue. It will prompt you to remove all ExpressCards, then ask for permission to load the PCI Express hotplug module (pciehp) if it is not loaded. PCIe hotplug capabilities are needed in order to add or remove PCIe-based ExpressCard cards while the system is running. Next the test will ask you for the number of ExpressCard slots in the system, followed by prompts to insert and remove cards with both types of interfaces (USB and PCIe) in any order.

The fv_core test is a wrapper that launches the FV guest and runs a core test on it. Please see Section A.1.1.12, “core” for information on the test methodology and run times.

The fv_memory test is a wrapper that launches the FV guest and runs a memory test on it. Please see Section A.1.1.26, “memory” for information on the test methodology and run times.

The fv_network test is a wrapper that launches the FV guest and runs a network test on it. It is useful for verifying the function of one or more network devices that support SR-IOV.

What the test covers: The test covers virtual function network devices on SR-IOV capable systems. Systems without SR-IOV may run the test too, but it will only verify the function of the standard virtual network hardware.

What the test does: Please see Section A.1.1.27, “network” for information on the test methodology and run times.

Preparing for the test: Assign a virtual function (VF) from a NIC to the guest. Directions on how to configure VFs can be found in the Using SR-IOV section of the Virtualization Deployment and Administration Guide.

Executing the test: The fv_network test is non-interactive. After properly assigning a VF to the guest, check the checkbox next to the test and click the Run Selected button to perform the test.

The fv_storage test is a wrapper that launches the FV guest and runs a storage test on it. It is not required for certification at this time.

What the test does: The Infiniband Connection test runs the following subtests to ensure a baseline functionality using, when appropriate, the ip address selected from the dropdown at the onset of the test:

Preparing for the test: Ensure that the LTS and SUT are separate machines, on the same fabric(s).

Executing the test: This is an interactive test. Check the box next to the test name to indicate it is among the tests to run. Click the button Run Selected to continue. You will be prompted with a dropdown to select an ip address (an ip address of LTS) in which to perform the tests using. Select an ip address corresponding to a device on the same fabric of the SUT device you are running the test for.

Manually adding and running the test: Use the following command to manually add the infinibandconnectiontest:

rhcert-cli plan --add --test infinibandconnectiontest --device <device name>_devicePort_<port number>

Use the following command to manually run the infinibandconnectiontest:

rhcert-cli run --test InfinibandConnectionTest --server <LTS IP addr>

Run time: This test takes less than 10 minutes to run.

What the test does: The info test is a part of all results packages. It’s run automatically along with any other test that is being performed and is a required part of every results package. If you attempt to submit a package that contains no info test, the package will be rejected. The test performs several different tasks. If any of these tasks fail, the info test fails:

After performing those tasks, the system gathers a sosreport and the output of dmidecode. These are used by our review team to help them in their analysis of the test results.

Run time: The info test takes around 1 minute on a 2013-era, single CPU, 3.3GHz, 6-core/12-thread Intel workstation with 8 GB of RAM running Red Hat Enterprise Linux 6.4, AMD64 and Intel 64 that was installed using the kickstart files in this guide. The time will vary depending on the speed of the machine and the quantity of RPM files that are installed.

What the test does: The IWarp Connection test runs the following subtests to ensure a baseline functionality using, when appropriate, the IP address selected from the dropdown at the onset of the test:

Preparing for the test: Ensure that the LTS and SUT are separate machines, on the same fabric(s).

Executing the test: This is an interactive test. Check the box next to the test name to indicate it is among the tests to run. Click the button Run Selected to continue. You will be prompted with a dropdown to select an ip address (an ip address of LTS) in which to perform the tests using. Select an ip address corresponding to a device on the same fabric of the SUT device you are running the test for.

Manually adding and running the test: Use the following command to manually add the iwarpconnectiontest:

rhcert-cli plan --add --test iWarpConnectionTest --device <device name>_devicePort_<port number>_netDevice_<net device here>

Use the following command to manually run the iwarpconnectiontest:

rhcert-cli run --test iWarpConnectionTest --server <LTS IP addr>

Run time: This test takes less than 10 minutes to run.

What the test covers: The kdump test verifies the ability of a system to capture a vmcore after a crash using the kdump utility. There are two entries in the local test plan, one for local core file storage and one for the remote copying of a vmcore via NFS to the test server.

What the test does: The test will crash the system and write a vmcore to /var/crash. It will crash the system a second time and write a vmcore to the /var/www/hwcert/export directory on the network / kdump server system. After each of the two actions occurs, the test server program will confirm that the system only did the things it was scheduled to do, e.g. it checks that only two reboots occurred when each panic was triggered.

Preparing for the test: Ensure that the system is connected to the network before running the test. All parameters will be automatically set by the test server.

Executing the test: The kdump test is interactive. Check the box next to the test name to indicate it is among the tests to run. Click the button Run Selected to continue. The system will ask you to click the Yes or No button to trigger the crash when the kdump test is run. The discs will sync and the vmcore file will be saved. You will see a series of messages including "Waiting for response", "Waiting for connection", and finally, "ready" as the test server waits for completion of the task. After the core is saved, the system under test will reboot and the rhcert application will be ready for the next test. The rhcert server will verify the vmcore file is present and valid. It will then repeat the crash, this time exporting the vmcore file to the test server, when you click the button next to the NFS version of tes test.

Run time: The kdump test run time is highly variable. It is dependent on the amount of RAM in the SUT, the speed of the disks in both the SUT and the test server, the speed of the network connection to the test server, and the time it takes to reboot the SUT. For a 2013-era workstation with 8GB of RAM, a 7200 RPM 6Gb/s SATA drive, a gigabit Ethernet connection to the test server and a 1.5 minute reboot time, a local kdump test can complete in about 4 minutes, including the reboot. The same 2013-era workstation can complete a NFS kdump test in about 5 minutes to a similarly equipped network test server. The required info test will add about a minute to the overall run time.

What the test covers: The lid test is only valid for systems that have integrated displays and therefore have a lid that can be opened and closed. The lid is detected by searching the udev database for a device with "lid" in its name:

E: NAME="Lid Switch"

What the test does: The test ensures that the system can determine when its lid is closed and when it is open via parameters in udev, and that it can turn off the display’s backlight when the lid is closed.

Preparing for the test: To prepare for the test, ensure that the power management settings do not put the system to sleep or into hibernation when the lid is closed. In Red Hat Enterprise Linux 6, right-click on the battery icon in the panel and choose Preferences. On the AC Power tab, select Blank screen as the action that occurs when the lid is closed. In Red Hat Enterprise Linux 7, use the Tweak Tool to disable suspend or hibernate on lid close. Make sure the lid is open before you start the test run.

Executing the test: The lid test is interactive. Check the box next to the test name to indicate it is among the tests to run. Click the button Run Selected to continue. You will be asked if you are ready to begin the test, so answer Yes to continue. Close the lid when prompted, watching to see if the backlight turns off. You may have to look through the small space between the keyboard and lid when the laptop is closed to verify that the backlight has turned off. Answer Yes if the backlight turns off or No if the backlight does not turn off.

Run time: The lid test takes about 30 seconds to perform, essentially the time it takes to close the lid just enough to have the backlight turn off. Because this test is run on laptops, a suspend test must accompany the required info test for each run. The suspend test will add approximately 6 minutes to each test run, and info will add another minute.

What the memory test covers: The memory test is used to test system RAM. It does not test USB flash memory, SSD storage devices or any other type of RAM-based hardware, it only tests main memory.

What the test does: The test uses the file /proc/meminfo to determine how much memory is installed in the system. Once it knows how much is installed, it checks to see if the system architecture is 32-bit or 64-bit. Then it determines if swap space is available or if there is no swap partition. The test runs either once or twice with slightly different settings depending on whether or not the system has a swap file:

In both iterations of the memory test, malloc() is used to allocate RAM, the RAM is dirtied with a write of an arbitrary hex string (0xDEADBEEF), and a test is performed to ensure that 0xDEADBEEF is actually stored in RAM at the expected addresses. The test calls free() to release RAM when testing is complete. Multiple threads or multiple processes will be used to allocate the RAM depending on whether the process size is greater than or less than the amount of memory to be tested.

Preparing for the test: Install the correct amount of RAM in the system in accordance with the rules in the Policy Guide.

Executing the test: The memory test is non-interactive. Check the checkbox next to the test and click the Run Selected button to perform the test.

Run time, bare-metal: The memory test takes about 16 minutes to run on a 2013-era, single CPU, 6-core/12-thread 3.3GHz Intel-based workstation with 8GB of RAM running Red Hat Enterprise Linux 6.4, AMD64 and Intel 64. The test will take longer on systems with more RAM. The required info test will add about a minute to the overall run time.

Run time, full-virt guest: The fv_memory test takes slightly longer than the bare-metal version, about 18 minutes, to run in a guest. The added time is due to guest startup/shutdown activities and the required info test that runs in the guest. The required info test on the bare-metal system will add about a minute to the overall run time. The fv_memory test run times will not vary as widely from machine to machine as the bare-metal memory tests, as the amount of RAM assigned to our pre-built guest is always the same. There will be variations caused by the speed of the underlying real system, but the amount of RAM in use during the test won’t change from machine to machine.

What the test covers: The network test is used to test devices whose function is transferring data over a network. This includes wired Ethernet cards, wireless Ethernet cards, virtual network devices on systems that support SR-IOV, and InfiniBand cards if IB is being used as a network protocol. The test will appear as network for non-Ethernet devices, or as different names for Ethernet or Wi-Fi devices:

What the test does: The test gathers information on all the network devices and runs this procedure:

Preparing for testing wired devices: You may test as many network devices from the official test plan as you wish in each run of the test suite. Connect each device at its native (maximum) speed or the test will fail. Ensure that the hwcert network test server is up and running before beginning, and make sure that each network device has an IP address assigned either statically or via DHCP.

If any network devices support partitioning, we need to see them demonstrate both full-speed data transfer and the partitioning function in one or more runs of the network test. This requirement will be accounted for in the official test plan by having two entries for each NIC that supports partitioning. If the NIC can run at full speed while it’s partitioned, please configure a partition with the NIC running at its native speed and perform your network tests in that configuration. This single test run will satisfy both official test plan entries for the NIC.

If the NIC cannot run at full speed while it’s partitioned, please perform one network test without partitioning so that we can see full-speed operation, and then perform another network test with partitioning enabled so that we can see a demonstration of the partitioning function. We recommend that you choose either 1Gb/s or 10Gb/s for your partitioned configuration so that it conforms to one of our existing network speed tests.

Preparing for testing wireless Ethernet devices: In Red Hat Enterprise Linux 6 and Red Hat Enterprise Linux 7, any system with a supported wireless card will automatically receive any necessary firmware package(s) at install time and all configuration of the cards can be done with the NetworkManager graphical tool. Simply select an SSID on a test network that does not require any additional user input during up/down operations (no authentication requests, VPN login, etc.) and you can run the test as explained in the "Executing the test" section below.

Executing the test: The network test is non-interactive. Check the checkbox next to the test and click the Run Selected button to perform the test.

Run time: The network test takes about 21 minutes for each PCIe-based, gigabit, wired Ethernet card that is being tested. We’ll add 10GbE test times and wireless times at a future date. The required info test will add about a minute to the overall run time.

What the test does: The Omnipath Connection test runs the following subtests to ensure a baseline functionality using, when appropriate, the ip address selected from the dropdown at the onset of the test:

Preparing for the test: Ensure that the LTS and SUT are separate machines, on the same fabric. You need to install opa-basic-tools on the LTS from the Downloads section of Red Hat customer portal web page.

Executing the test: This is an interactive test. Check the box next to the test name to indicate it is among the tests to run. Click the button Run Selected to continue. You will be prompted with a dropdown to select an ip address (an ip address of LTS) in which to perform the tests using. Select an ip address corresponding to a device on the same fabric of the SUT device you are running the test for.

Manually adding and running the test: Use the following command to manually add the OminpathConnectionTest:

rhcert-cli plan --add --test ominpathconnectiontest --device <device name>_devicePort_<port number>

Use the following command to manually run the OmnipathConnectionTest:

rhcert-cli run --test ominpathconnectiontest --server <LTS IP addr>

Run time: This test takes less than 10 minutes to run.

What the test covers: The pccard test covers PC Cards (also known as PCMCIA cards).

What the test does: The test uses the /sbin/pccardctl command to control the system’s pccard sockets individually. It loops through all the sockets and performs three actions: a power off, power on and a card query to get the identity of the inserted card(s).

Preparing for the test: Each card slot must be populated before running the test. The /sbin/pccardctl utility has the ability to turn the slots off and on, simulating an eject and an insert, so the tester is not prompted to insert cards at test time.

Executing the test: The pccard test is non-interactive. Check the checkbox next to the test and click the Run Selected button to perform the test.

What the test does: The profiler test will attempt to shut down oprofile to get a clean slate, then bring it up correctly (start the daemon). It will load all the oprofile modules and a handful of additional support items (for example, some directories under /dev are mounted), then it will start the oprofile application. The application will acquire some sample data, called a report, then quit. If all those steps are completed successfully, the test passes. There is another loop in the test that is executed if one of those actions fails. The oprofile application requires specific hardware registers in the CPU to record its data. If for some reason this dedicated support is not working (or the hardware counters are not present), the other loop enables timer mode, allowing the data to be recorded in software instead of in the CPU registers. If you encounter failures in the profiler test, try forcing timer mode by adding this line to /etc/modprobe.conf and then rebooting before attempting to run the test again:

options oprofile timer=1

Preparing for the test: To prepare for the test, ensure that a CPU is installed that meets the requirements explained in the Policy Guide.

Executing the test: The profiler test is non-interactive. Check the checkbox next to the test and click the Run Selected button to perform the test.

Run time: The profiler test takes approximately 30 seconds on a 2013-era workstation. The required info test will add about a minute to the overall run time.

What the test covers: The realtime test covers the testing of systems running Red Hat Enterprise Linux for Real Time with two sets of tests: one to find system management mode-based execution delays, and one to determine the latency of servicing timer events.

What the test does: The first portion of the test loads a special kernel module named hwlat_detector.ko. This module creates a kernel thread which polls the Timestamp Counter Register (TSC), looking for intervals between consecutive reads which exceed a specified threshold. Gaps in consecutive TSC reads mean that the system was interrupted between the reads and executed other code, usually System Management Mode (SMM) code defined by the system BIOS.

The second part of the test starts a program named cyclictest, which starts a measurement thread per cpu, running at a high realtime priority. These threads have a period (100 microseconds) where they perform the following calculation:

The latency is the time difference between the theoretical wakeup time (t1+period) and the actual wakeup time (t2). Each measurement thread tracks minimum, maximum and average latency as well as reporting each datapoint.

Once cyclictest is running, rteval starts a pair of system loads, one being a parallel linux kernel compile and the other being a scheduler benchmark called hackbench.

When the run is complete, rteval performs a statistical analysis of the data points, calculating mean, mode, median, variance and standard deviation.

Preparing for the test: Install and boot the realtime kernel-rt kernel before adding the system to the certification. The command will detect that the running kernel is a realtime kernel and will schedule the realtime test to be run.

Running the test: The realtime test is non-interactive. Check the checkbox next to the test and click the Run Selected button to perform the test. The test will only appear when the system is running the rt-kernel.

Run time: The system management mode portion of the test runs for two hours. The timer event analysis portion of the test runs for twelve hours on all machines. The required info test will add about a minute to the overall run time.

What the test covers: The reboot test confirms the ability of a system to reboot when prompted. It is not required for certification at this time.

What the test does: The test issues a shutdown -r 0 command to immediately reboot the system with no delay.

Preparing for the test: Ensure that the system can be rebooted before running this test by closing any running applications.

Executing the test: The reboot test is interactive. Check the box next to the test name to indicate it is among the tests to run. Click the button Run Selected to continue. You will be asked Ready to restart? when you reach the reboot portion of the test program. Answer y if you are ready to perform the test. The system will reboot and after coming back up, the test server will verify that the reboot completed successfully.

What the test does: The Omnipath Connection test runs the following subtests to ensure a baseline functionality using, when appropriate, the ip address selected from the dropdown at the onset of the test:

Preparing for the test: Ensure that the LTS and SUT are separate machines, on the same fabric(s).

Executing the test: This is an interactive test. Check the box next to the test name to indicate it is among the tests to run. Click the button Run Selected to continue. You will be prompted with a dropdown to select an ip address (an ip address of LTS) in which to perform the tests using. Select an ip address corresponding to a device on the same fabric of the SUT device you are running the test for.

Manually adding and running the test: Use the following command to manually add the RoCEConnectionTest:

rhcert-cli plan --add --test RoCEConnectionTest --device <device name>_devicePort_<port number>_netDevice_<net device here>

Use the following command to manually run the RoCEConnectionTest:

rhcert-cli run --test RoCEConnectionTest --server <LTS IP addr>

Run time: This test takes less than 10 minutes to run.

What the SATA test covers:

There are many different kinds of persistent on-line storage devices available in systems today.

What the test does:

The SATA test is designed to test anything that reports an ID_TYPE of "disk" in the udev database. This test is for SATA drives. The hwcert/storage/SATA test gets planned if:

If the above two criteria do not meet, then the storage test would get planned for the detected device.

For more information on what the test does and preparing for the test see Section A.1.1.44, “STORAGE”

What the SATA_SSD test covers:

This test will run if it determines the storage unit of interest is SSD and its interface is SATA.

What the SATA_SSD test does:

The test finds the SCSI storage type and identifies connected storage interface on the location more /sys/block/sdap/queue/rotational. The test is planned if the rotational bit is set to zero for SSD.

Following are the device parameter values that would be printed as part of the test:

For more information on what the test does and preparing for the test see Section A.1.1.44, “STORAGE”

What the M2_SATA test covers:

This test will run if it determines the interface is SATA and attached through an M2 connection.

Manually adding and running the test:

Use the following command to manually add the M2_SATA test:

rhcert-cli plan --add --test M2_SATA --device host0

Following are the device parameter values that would be printed as part of the test:

For more information on what the test does and preparing for the test see Section A.1.1.44, “STORAGE”

What the U2_SATA test covers:

This test will run if it determines the interface is SATA and attached through a U2 connection.

Manually adding and running the test:

Use the following command to manually add the U2_SATA test:

rhcert-cli plan --add --test U2_SATA --device host0

Following are the device parameter values that would be printed as part of the test:

For more information on what the test does and preparing for the test see Section A.1.1.44, “STORAGE”

What the SAS test covers:

There are many different kinds of persistent on-line storage devices available in systems today.

What the test does:

The SAS test is designed to test anything that reports an ID_TYPE of "disk" in the udev database. This test is for SAS drives. The hwcert/storage/SAS test gets planned if:

If the above two criteria do not meet, then the storage test would get planned for the detected device.

For more information on what the test does and preparing for the test see Section A.1.1.44, “STORAGE”

What the SAS_SSD test covers:

This test will run if it determines the storage unit of interest is SSD and its interface is SAS.

What the SAS_SSD test does:

The test finds the SCSI storage type and identifies connected storage interface on the location more /sys/block/sdap/queue/rotational. The test is planned if the rotational bit is set to zero for SSD.

Following are the device parameter values that are printed as part of the test:

For more information on what the test does and preparing for the test see Section A.1.1.44, “STORAGE”

What the PCIe_NVMe test covers:

This test will run if it determines the interface is NVMe and attached through a PCIE connection.

What the PCIe_NVMe test does:

This test gets planned if logical device host name string contains " nvme[0-9] "

Following are the device parameter values that would be printed as part of the test:

For more information on what the test does and preparing for the test see Section A.1.1.44, “STORAGE”

What the M2_NVMe test covers:

This test will run if it determines the interface is NVMe and attached through an M2 connection.

Manually adding and running the test:

Use the following command to manually add the M2_NVMe test:

rhcert-cli plan --add --test M2_NVMe --device nvme0

Following are the device parameter values that would be printed as part of the test:

For more information on what the test does and preparing for the test see Section A.1.1.44, “STORAGE”

What the U2_NVMe test covers:

This test will run if it determines the interface is NVMe and attached through a U2 connection.

Manually adding and running the test:

Use the following command to manually add the U2_NVMe test:

rhcert-cli plan --add --test U2_NVMe --device nvme0

Following are the device parameter values that would be printed as part of the test:

For more information on what the test does and preparing for the test see Section A.1.1.44, “STORAGE”

[[sect- NVDIMM]]

What the NVDIMM test covers:

This test operates like any other SSD non-rotational storage test and identifies the NVDIMM storage devices

What the test does:

The test gets planned for storage device if:

Following are the device parameter values that would be printed as part of the test:

For more information on what the test does and preparing for the test see Section A.1.1.44, “STORAGE”

What the storage test covers: There are many different kinds of persistent on-line storage devices available in systems today. The STORAGE test is designed to test anything that reports an ID_TYPE of "disk" in the udev database. This includes IDE, SCSI, SATA, SAS, and SSD drives, PCIe SSD block storage devices, as well as SD media, xD media, MemoryStick and MMC cards. The test plan script reads through the udev database and looks for storage devices that meet the above criteria. When it finds one, it records the device and its parent and compares it to the parents of any other recorded devices. It does this to ensure that only devices with unique parents are tested. If the parent has not been seen before, the device is added to the test plan. This speeds up testing as only one device per controller will be tested, as per the Policy Guide.

What the test does: The STORAGE test performs the following actions on all storage devices with a unique parent:

Preparing for the test: You should install all the drives and storage controllers that are listed on the official test plan. In the case of multiple storage options, as many as can fit into the system at one time can be tested in a single run, or each storage device can be installed individually and have its own run of the storage test. You can decide on the order of testing and number of controllers present for each test. Each logical drive attached to the system must contain a swap partition in addition to any other partitions, or be totally blank. This is to provide the test with a location to create a filesystem and run the tests. The use of swap partitions will lead to a much quicker test, as devices left blank are tested in their entirety. They will almost always be significantly larger than a swap partition placed on the drive. Please see the Red Hat Knowledgebase article at https://access.redhat.com/site/solutions/15244 for more information on appropriate swap file sizing.

When it comes to choosing storage devices for the official test plan, the rule that the review team operates by is "one test per code path". What we mean by that is that we want to see a storage test run using every driver that a controller can use. The scenario of multiple drivers for the same controller usually involves RAID storage of some type. It’s common for storage controllers to use one driver when in regular disk mode and another when in RAID mode. Some even use multiple drivers depending on the RAID mode that they are in. The review team will analyze all storage hardware to determine the drivers that need to be used in order to fulfill all the testing requirements. That’s why you may see the same storage device listed more than once in the official test plan. Complete information on storage device testing is available in the Policy Guide.

Executing the test: The storage test is non-interactive. Check the checkbox next to the test and click the Run Selected button to perform the test.

Host bus adapter host0 has storage devices sda, sda1, sda2, sda3
Which disk would you like to test:  (sda|sda1|sda2|sda3|all)

Run time, bare-metal: The storage test takes approximately 22 minutes on a 6Gb/s SATA hard drive installed in a 2013-era workstation system. The same test takes approximately 3 minutes on a 6Gb/s SATA solid-state drive installed in a 2013-era workstation system. The required info test will add about a minute to the overall run time.

What the test covers: The suspend test covers suspend/resume from S3 sleep state (suspend to RAM) and suspend/resume from S4 hibernation (suspend to disk). This test is only scheduled on systems that have built-in batteries, like laptops, so it won’t be present on any other type of system.

What the test does: The test queries the /sys/power/state file and determines which states are supported by the hardware. If it sees "mem" in the file, it schedules the S3 sleep test. If it sees "disk" in the file, it schedules the S4 hibernation test. If it sees both, it schedules both. What follows is the procedure for a system that supports both S3 and S4 states. If your system does not support both types it will only run the tests related to the supported type.

Preparing for the test: Ensure that a swap file large enough to hold the contents of RAM was created when the system was installed. Guidelines for swap file size can be found at this Red Hat Knowledgebase article: https://access.redhat.com/site/solutions/15244. Also, someone must be present at the system under test in order to wake it up from suspend and hibernate.

Executing the test: The suspend test is interactive. Check the box next to the test name to indicate it is among the tests to run. Click the button Run Selected to continue. The test server GUI will display a status of suspend? when the test runs. Click on the suspend? status link or the Continue Testing button and then click the Yes button to suspend the laptop.

The test server will display waiting for response after it sends the suspend command. Check the laptop and confirm that it has completed suspending, then press the power button or any other key that will wake it from suspend. The test server will continuously monitor the system under test to see if it has awakened. Once it has woken up, the test server GUI will display the question Has resume completed?. Press the Yes or No button to tell the test server what happened.

The server will then continue to the hibernate test. Again, click the Yes button under the suspend? question to put the laptop into hibernate mode.

The test server will display waiting for response after it sends the hibernate command. Check the laptop and confirm that it has completed hibernating, then press the power button or any other key that will wake it from hibernation. The test server will continuously monitor the system under test to see if it has awakened. Once it has woken up, the test server GUI will display the question has resume completed?. Press the Yes or No button to tell the test server what happened.

Next the test server will ask you if the system has a keyboard key that will cause the system under test to suspend. If it does, click the Yes button under the question Does this system have a function key (Fn) to suspend the system to mem?. Follow the procedure described above to verify suspend and wake the system up to continue with testing.

Finally the test server will ask you if the system has a keyboard key that will cause the system under test to hibernate. If it does, click the Yes button under the question Does this system have a function key (Fn) to suspend the system to disk? Follow the procedure described above to verify hibernation and wake the system up to continue with any additional tests you have scheduled.

Run time: The suspend test takes about 6 minutes on a 2012-era laptop with 4GB of RAM and a non-SSD hard drive. This is the time for a full series of tests, including both pm-suspend-based and function-key-based suspend and hibernate runs. The time will vary depending on the speed at which the laptop can write to disk, the amount and speed of the RAM installed, and the capability of the laptop to enter suspend and hibernate states through function keys. The required info test will add about a minute to the overall run time.

What the test covers: The tape test covers all types of tape drives. Any robots associated with the drives are not tested by this test.

What the test does: The test uses the mt command to rewind the tape, then it does a tar of the /usr directory and stores it on the tape. A tar compare is used to determine if the data on the tape matches the data on the disk. If the data matches, the test passes.

Preparing for the test: Insert a tape of the appropriate size into the drive.

Executing the test: The tape test is non-interactive. Check the checkbox next to the test and click the Run Selected button to perform the test.

What the test covers: The USB2 test covers USB2 ports from a basic functionality standpoint, ensuring that all ports can be accessed by the OS.

What the test does: The purpose of the test is to ensure that all USB2 ports present in a system function as expected. It asks for the number of available USB2 ports (minus any that are in use for keyboard/mouse, etc.) and then asks the tester to plug and unplug a USB2 device into each port. The test watches for attach and detach events and records them. If it detects both plug and unplug events for the number of unique ports the tester entered, the test will pass.

Preparing for the test: Count the available USB2 ports and have a spare USB2 device available to use during the test. You may need to trace the USB ports from the motherboard header(s) to distinguish between USB2 and USB3 ports.

Executing the test: The USB2 test is interactive. Check the box next to the test name to indicate it is among the tests to run. Click the button Run Selected to continue. When prompted by the system, enter the number of available USB2 ports present on the system. Don’t count any that are currently in use by keyboards or mice. The system will ask for the test USB2 device to be plugged into a port and will then pause until the tester presses y to continue. The system will then ask for the device to be unplugged and again will pause until the tester presses y to continue. These steps repeat for the number of ports that were entered. Note that there is no right or wrong order for testing the ports, but each port must be tested only once.

Run time: The USB2 test takes about 15 seconds per USB2 port. This includes the time to manually plug in the device, scan the port, unplug the device, and scan the port again. The required info test will add about a minute to the overall run time.

What the test covers: The USB3 test covers USB3 ports from a basic functionality standpoint, ensuring that all ports can be accessed by the OS.

What the test does: The purpose of the test is to ensure that all USB3 ports present in a system function as expected. It asks for the number of available USB3 ports (minus any that are in use for keyboard/mouse, etc.) and then asks the tester to plug and unplug a USB3 device into each port. The test watches for attach and detach events and records them. If it detects both plug and unplug events for the number of unique ports the tester entered, the test will pass.

Preparing for the test: Count the available USB3 ports and have a spare USB3 device available to use during the test. You may need to trace the USB ports from the motherboard header(s) to distinguish between USB2 and USB3 ports.

Executing the test: The USB3 test is interactive.Check the box next to the test name to indicate it is among the tests to run. Click the button Run Selected to continue. When prompted by the system, enter the number of available USB3 ports present on the system. Don’t count any that are currently in use by keyboards or mice. The system will ask for the test USB3 device to be plugged into a port and will then pause until the tester presses y to continue. The system will then ask for the device to be unplugged and again will pause until the tester presses y to continue. These steps repeat for the number of ports that were entered. Note that there is no right or wrong order for testing the ports, but each port must be tested only once.

Run time: The USB3 test takes about 15 seconds per USB3 port. This includes the time to manually plug in the device, scan the port, unplug the device, and scan the port again. The required info test will add about a minute to the overall run time.

What the test covers: All video hardware, whether removeable or integrated on the motherboard, is tested using the video test. Devices are selected for testing by their PCI class ID. Specifically, the test is looking for a device class of "30000" in the output of udev.

What the test does: The video test first determines which command is used to control the X configuration on the machine where it is running (either redhat-config-xfree86 or system-config-display). It then runs it with the --noui flag and generates a clean X configuration file. It runs startx using the new configuration file and runs x11perf, which is a X11 server performance test program. After the performance test completes it also runs xdpyinfo to determine the screen resolution and color depth. The configuration file created at the start of the test should allow the system to run at the maximum resolution that the monitor and video card are capable of achieving. The final potion of the test uses grep to search through the /var/log/Xorg.0.log logfile to determine which driver is being used.

Preparing for the test: Ensure that the monitor and video card in the system are capable of running at a resolution of 1024x768 with a color depth of 24 bits per pixel (bpp). This is the minimum resolution and color depth required to achieve a passing video test. Higher resolutions or color depths are also acceptable, but nothing lower than 1024x768 at 24bpp will pass. You can confirm this ability by looking at the output of xrandr. All the resolutions that can be achieved by the monitor and video card should be displayed in the output of xrandr. Check the output for 1024x768 at 24 bits per pixel (or higher). You may need to remove any KVM switches that are between the monitor and video card if you are not seeing all the resolutions that the card/monitor combination are capable of generating.

Executing the test: The video test is non-interactive. Check the checkbox next to the test and click the Run Selected button to perform the test. The screen on the test system will go blank, followed by a series of test patterns from the x11perf test program. It will return to the desktop or to the virtual terminal screen that the system was on at execution time when the test finishes.

Run time: The video test takes about 1 minute to perform on a 2013-era workstation. The required info test will add about a minute to the overall run time.

What the test covers: The WirelessG test is run on all wireless Ethernet connections with a maximum connection speed of 802.11g.

What the test does: This is a new test that combines the existing wlan and network tests. In addition to passing all the existing network test items, this test must detect a "g" link type as reported by iw and demonstrate a throughput of 22Mb/s (with a margin for overhead) in order to pass. Please see Section A.1.1.27, “network” for information on the rest of the test functionality.

What the test covers: The WirelessN test is run on all wireless Ethernet connections with a maximum connection speed of 802.11n.

What the test does: This is a new test that combines the existing wlan and network tests. In addition to passing all the existing network test items, this test must detect an "n" link type as reported by iw and demonstrate a throughput of 100Mb/s (with a margin for overhead) in order to pass. Please see Section A.1.1.27, “network” for information on the rest of the test functionality.

What the test covers: The WirelessAC test is run on all wireless Ethernet connections with a maximum connection speed of 802.11ac.

What the test does: This is a new test that combines the existing wlan and network tests. In addition to passing all the existing network test items, this test must detect an "ac" link type as reported by iw and demonstrate a throughput of 300Mb/s (with a margin for overhead) in order to pass. Please see Section A.1.1.27, “network” for information on the rest of the test functionality.