MAP 0050: SCSI bus problems

Use this MAP to analyze problems with a SCSI bus.

Purpose of this MAP

Use this MAP to analyze problems with a SCSI bus.

Considerations

Follow the steps in this MAP to isolate a SCSI bus problem.

Note:
This procedure steps you through a process to systematically remove devices and components from a SCSI bus until a problem or a symptom or problem is eliminated. If you go through the whole procedure you will remove all components of a SCSI bus in the following order:
  1. Hot-swap devices
  2. Devices that are not hot-swap
  3. SCSI Enclosure Services (SES) device or enclosures
  4. SCSI cables
  5. SCSI adapter

Do the following:

Preparing for hot-plug SCSI device or cable deconfiguration

Use this procedure when you are preparing to deconfigure a hot-plug SCSI device or cable. This procedure will help determine if a SCSI device or SCSI device cable is causing your system problem.

Purpose of this procedure

Use this procedure to determine if a SCSI device or SCSI device cable is causing the symptom.

Disconnect all cables attached to the adapter, (except for the cable to the device from which you boot to run diagnostics; you may want to temporarily move this device to another SCSI port while you are trying to find the problem).

Note:
Move this device to another SCSI port while you are trying to find the problem.

  1. Go to Running the online and eServer stand-alone diagnostics and perform the prerequisite tasks described in the "Before you begin" topic.
  2. Determine which SCSI adapter you plan to remove the cables or devices from.
  3. Adapter slots are numbered on the rear of the system unit, record the slot number and location of each adapter being removed.
  4. Ensure that any processes or applications that might use the adapter are stopped.
  5. Enter the system diagnostics by logging in as "root" user or as the "celogin" user. Type the diag command on the AIX(R) command line.
  6. When the "DIAGNOSTIC OPERATING INSTRUCTIONS" menu displays, press Enter. The "FUNCTION SELECTION menu appears.
  7. From the "FUNCTION SELECTION" menu, select "Task Selection", then press Enter.
  8. From the "Task Selection" list, select "PCI Hot Plug Manager".
  9. From the PCI Hot Plug Manager menu, select "Unconfigure a Device", then press Enter.
  10. Press F4 or ESC 4 to display the "Device Names" menu.
  11. Select the adapter from which you are removing the cables or devices in the "Device Names" menu.
  12. In the "Keep Definition" field, use the tab key to answer "Yes".
  13. In the "Unconfigure Child Devices" field, use the Tab key to answer "Yes", then press Enter.
  14. The "ARE YOU SURE" screen displays. Press Enter to verify the information. A successful deconfiguration is indicated by the "OK" message displayed next to the Command field at the top of the screen.
  15. Press F4 or ESC 4 twice to return to the "Hot Plug Manager" menu.
  16. Select "replace/remove PCI Hot Plug adapter".
  17. Select the slot that has the adapter you want to remove the cables or devices from in the system.
  18. Select "remove".
    Note:
    A fast blinking amber LED located at the back of the machine near the adapter indicates that the slot has been identified.
  19. Press Enter. This places the adapter in the "action" state, meaning it is ready to be removed from the system. (Don't need to remove the adapter, unless it makes removing the cables attached to it easier).

After hot-plug SCSI device or cable deconfiguration

Use this procedure after you deconfigure or hot-plug a SCSI device to ensure that the replaced component was successfully installed.

  1. Press Enter, then continue to follow the screen instructions until you receive a message that the replacement is successful. A successful replacement is indicated by the OK message displayed next to the command field at the top of the screen.
  2. Press the F3 or ESC 3 key to return to the PCI Hot-Plug Manager menu.
  3. Press the F3 or ESC 3 key to return to the TASK selection list.
  4. Select "Log Repair Action".
  5. Select the adapter you just removed the cables or devices from, then press Enter.
  6. Press Commit (F7 or ESC 7), then press Enter.
  7. Press the F10 or the ESC 0 key to exit diagnostics.
  8. Type the diag -a command on the command line.

SCSI service hints

Use one or more of the following procedures when servicing SCSI adapter or devices.

General SCSI Configuration Checks

The following steps apply to all types of SCSI problems:

  1. Verify that all SCSI devices on the SCSI bus have a unique address.
  2. Verify that all cables are connected securely and that there is proper termination at both ends of the SCSI bus.
  3. Verify that the cabling configuration does not exceed the maximum cable length for the adapter in use.
  4. Verify that the adapters and devices that you are working with are at the appropriate microcode levels for the customer situation. If you need assistance with microcode issues, contact your service support structure.
  5. If there are multiple SCSI adapters on the SCSI bus, verify that the customer is using the appropriate software (such as HACMP(TM) or HANFS) to support such an arrangement. If the correct software is not in use, some SCSI errors should be expected when multiple adapters attempt to access the same SCSI device. Also, each adapter should have a unique address.
High Availability or Multiple SCSI System Checks

If you have a high-availability configuration, or if more than one system is attached to the same SCSI bus, do the following:

  1. Verify that the adapters and devices have unique SCSI addresses. The default SCSI adapter address is always 7. If you have more than one adapter on the bus, change the address of at least one adapter. This can be done by using SMIT (SMIT Devices -> SCSI Adapter -> Change/Show characteristics of an adapter). You must make the changes to the database only, then reboot the system in order for the change to take effect.
    Note:
    Diagnostics defaults to using ID 7 (it is recommended that this ID not be used in high availability configurations).
  2. If RAID devices such as the 7135 or 7137 are attached, be sure to run the proper diagnostics for the device. If problems occur, contact your service support structure for assistance. If the diagnostics are run incorrectly on these devices, misleading SRNs can result.
  3. Diagnostics cannot be run against OEM devices; doing so results in misleading SRNs.
  4. Verify that all cables are connected securely and that both ends of the SCSI bus is terminated correctly.
  5. Verify that the cabling configuration does not exceed the maximum cable length for the adapter in use. Refer to the SCSI Cabling section in the RS/6000(R) eServer(TM) pSeries(R) Adapters, Devices, and Cable Information for Multiple Bus Systems for more details on SCSI cabling issues.
  6. Verify that adapter and devices are at the appropriate microcode levels for the customer situation. If you need assistance with microcode issues, contact your service support structure.
SCSI-2 Single-Ended Adapter PTC Failure Isolation Procedure

Before replacing a SCSI-2 single-ended adapter, use these procedures to determine if a short-circuit condition exists on the SCSI bus. The same positive temperature coefficient (PTC) resistor is used for both the internal and external buses. The PTC protects the SCSI bus from high currents due to shorts on the cable, terminator, or device. It is unlikely that the PTC can be tripped by a defective adapter. Unless instructed to do so by these procedures, do not replace the adapter because of a tripped PTC resistor.

A fault (short-circuit) causes an increase in PTC resistance and temperature. The increase in resistance causes the PTC to halt current flow. The PTC returns to a low resistive and low temperature state when the fault is removed from the SCSI bus or when the system is turned off. Wait 5 minutes for the PTC resistor to fully cool, then retest.

These procedures determine if the PTC resistor is still tripped and then determine if there is a short somewhere on the SCSI bus.

Determining Where to Start

Use the following to determine the adapter configuration and select the proper procedure:

External SCSI-2 Single-Ended Bus PTC Isolation Procedure

Isolate the external SCSI bus PTC fault with the following procedure:

Note:
The external bus is of single-ended design.
  1. Ensure the system power and all externally attached device power is turned off. All testing is accomplished with the power off.
  2. Disconnect any internal and external cables from the adapter and remove the adapter from the system.
  3. Verify with a digital Ohmmeter that the internal PTC resistor, labeled Z1, (refer to the illustration after Internal SCSI-2 Single-Ended Bus PTC Isolation Procedure, step 3) is cool and in a low resistance state, typically less than 1/2 Ohm. Measuring across, be sure to probe both sides of the PTC where the solder joints and board come together. The polarity of the test leads is not important. If necessary, allow the PTC resistor to cool and measure again.
  4. This step determines if there is a short on the adapter. Locate Capacitor C1 and measure the resistance across it by using the following procedure:
    1. Connect the positive lead to the side of the capacitor where the + is indicated on the board near C1. Be sure to probe at the solder joint where the capacitor and board come together.
    2. Connect the negative lead to the opposite side of the capacitor marked "GND." Be sure to probe at the solder joint where the capacitor and board come together.
    3. If there is no short present, then the resistance reading is high, typically hundreds of Ohms.
    Note:
    Because this is a measurement across unpowered silicon devices, the reading is a function of the Ohmmeter used.
  5. Reattach the external cable to the adapter, then do the following:
    1. Measure across C1 as previously described.
    2. If the resistance is still high, in this case above 10 Ohms, then there is no apparent cause for a PTC failure from this bus. If there are internal cables attached continue to the Internal SCSI-2 Single-Ended Bus PTC Isolation Procedure.
    3. If the resistance is less than 10 Ohms, there is a possibility of a fault on the external SCSI bus. Troubleshoot the external SCSI bus by disconnecting devices and terminators. Measure across C1 to determine if the fault has been removed. Replace the failing component. Go to Verifying a repair.
External SCSI-2 Single-Ended Bus Probable Tripped PTC Causes

The following list provides some suggestions of things to check when the PTC is tripped:

Internal SCSI-2 Single-Ended Bus PTC Isolation Procedure

Isolate the internal SCSI bus PTC resistor fault with the following procedure:

Note:
The internal bus is single-ended.
  1. Ensure that system power and all externally attached device power is turned off.
  2. Disconnect any internal and external cables from the adapter then remove the adapter from the system.
  3. Verify with a digital Ohmmeter, that the internal PTC resistor, labeled Z1, is cool and in a low resistance state, typically less than 1/2 Ohm. Measuring across, be sure to probe both sides of the PTC where the solder joints and board come together. The polarity of the test leads is not important. If necessary, allow the PTC to cool and measure again. Refer to the following illustration.
    a4aa0148
    Note:
    Only the probe tips are touching the solder joints. Do not allow the probes to touch any other part of the component.
  4. This step determines if there is a short on the adapter. Locate capacitor C1 and measure the resistance across it using the following procedure:
    1. Connect the positive lead to the side of the capacitor where the + is indicated. Be sure to probe at the solder joint where the capacitor and board come together.
    2. Connect the negative lead to the opposite side of the capacitor. Be sure to probe at the solder joint where the capacitor and board come together.
    3. If there is no short present, the resistance reading is high, typically hundreds of Ohms.
    Note:
    Because this is a measurement across unpowered silicon devices, the reading is a function of the Ohmmeter used.
  5. Reattach the internal cable to the adapter, then do the following:
    1. Measure across C1 as described above.
    2. If the resistance is still high, above 25 Ohms, there is no apparent cause for a PTC failure.
    3. If the resistance is less than 10 Ohms, a fault on the internal SCSI bus is possible. Troubleshoot the internal SCSI bus by disconnecting devices and terminators. Measure across C1 to determine if the fault has been removed.
      Note:
      Some internal cables have nonremovable terminators.
Internal SCSI-2 Single-Ended Bus Probable Tripped PTC Resistor Causes

The following list provides some suggestions of things to check when the PTC is tripped:

SCSI-2 Differential Adapter PTC Failure Isolation Procedure

Use this procedure when SRN xxx-240 or xxx-800 has been indicated.

The differential adapter can be identified by the 4-B or 4-L on the external bracket plate.

Before replacing a SCSI-2 differential adapter, use these procedures to determine if a short-circuit condition exists on the SCSI Bus. The PTC protects the SCSI bus from high currents due to shorts on the cable, terminator, or device. It is unlikely that the PTC can be tripped by a defective adapter. Unless instructed to do so by these procedures, do not replace the adapter because of a tripped PTC resistor.

A fault (short-circuit) causes an increase in PTC resistance and temperature. The increase in resistance causes the PTC to halt current flow. The PTC returns to a low resistive and low temperature state when the fault is removed from the SCSI bus or when the system is turned off. Wait 5 minutes for the PTC resistor to fully cool, then retest.

These procedures determine if the PTC resistor is still tripped and then determine if there is a short somewhere on the SCSI bus.

External SCSI-2 Differential Adapter Bus PTC Isolation Procedure

Isolate the external SCSI bus PTC fault with the following procedure:

Notes:
  1. Only the probe tips are touching the solder joints. Do not allow the probes to touch any other part of the component.
  2. The external bus is differential.
  1. Ensure that system power and all externally attached device power is turned off.
  2. Check to ensure all devices are marked SCSI Differential and that the terminator on the end of the SCSI bus is also marked differential. If not, you may have a single-ended SCSI device or terminator on the differential SCSI bus. Single-ended devices do not work on a differential SCSI bus and may cause a PTC type error to be reported. The entire SCSI bus may appear to be intermittent. After ensuring the system is completely differential, continue.
  3. Disconnect the external cables from the adapter and remove the adapter from the system.
  4. Verify with a digital Ohmmeter that the internal PTC resistor, labeled Z1, (refer to the illustration on page External SCSI-2 Differential Adapter Bus PTC Isolation Procedure) is cool and in a low resistance state, typically less than 1/2 Ohm. Measuring across, be sure to probe both sides of the PTC resistor where the solder joints and board come together. The polarity of the test leads is not important. If necessary, allow the PTC resistor to cool and measure again.
    a4aa0149
  5. This step determines if there is a short on the adapter. Locate capacitor C1 and measure the resistance across it using the following procedure:
    1. Connect the negative lead to the side of the capacitor marked "GND". Be sure to probe at the solder joint where the capacitor and board come together.
    2. Connect the positive lead to the side of the capacitor marked "Cathode D1" on the board near C1. Be sure to probe at the solder joint where the capacitor and board come together.
      • If there is no fault present, then the resistance reading is 25 to 35 Ohms. The adapter is not faulty. Continue to the next step.
      • If the resistance measured is higher than 35 Ohms, check to see if RN1, RN2, and RN3 are plugged into their sockets. If these sockets are empty, you are working with a Multi-Initiators or High-Availability system. With these sockets empty, a resistive reading across C1 cannot be verified other than it measures a high resistance (not a short). If the resistance measurement is not low enough to be suspected as a fault (lower than 10 Ohms), continue to the next step.
      • If the resistance is high and there is no external cable to reattach, there is no apparent cause for the PTC diagnostic failure.
      • If the resistance reading is low, typically below 10 Ohms, there is a fault. Because there are no cables attached, the fault is on the adapter. Replace the adapter.
      • If the resistance measured was high and there is an external cable to reattach, proceed to the next step.
  6. Reattach the external cable to the adapter.
    1. Measure across C1 as previously described.
    2. If the resistance is between 10 to 20 Ohms, there is no apparent cause for a PTC resistor failure.
    3. If the resistance is less than 10 Ohms, there is a possibility of a fault on the external SCSI bus. Troubleshoot the external SCSI bus by disconnecting devices and terminators. Measure across C1 to determine if the fault has been removed.
SCSI-2 Differential Adapter Probable Tripped PTC Causes

The following list provides some suggestions of things to check when the PTC is tripped:

Dual-Channel Ultra SCSI Adapter PTC Failure Isolation Procedure

Use the following procedures if diagnostics testing indicates a potential positive temperature coefficient (PTC) resistor fault or the TERMPWR Shorted LED is lit.

This procedure is used for SRNs 637-240 and 637-800 on the Dual-Channel Ultra SCSI Adapter. If the TERMPWR Shorted LED is lit, use this procedure to help isolate the source of the problem on the failing channel.

  1. Identify the adapter by its label of 4-R on the external bracket. Then, determine if the failure is on channel A or channel B.
  2. The same PTC is used for both the internal and external buses. The PTC protects the SCSI bus from high currents due to shorts on the cable, terminator, or device. It is unlikely that the PTC can be tripped by a defective adapter. A fault (short-circuit) causes an increase in PTC resistance and temperature. The increase in resistance causes the PTC to halt current flow. The PTC returns to a low resistive and low temperature state when the fault is removed from the SCSI bus or when the system is turned off.

    Wait 5 minutes for the PTC resistor to fully cool, then retest.

  3. If this same error persists, or the TERMPWR Shorted LED is lit, replace the components of the failing channel in the following order (wait five minutes between steps):
    1. If the failure is on the external cable, replace the following:
      1. Cable
      2. Device
      3. Attached subsystem
      4. Adapter
    2. If the failure is on the internal cable, replace the following:
      1. Cable
      2. Device
      3. Backplane
      4. Adapter
    3. If the failure persists, verify that the parts exchanged are in the correct channel (internal or external, A or B).

      If the errors are still occurring, continue isolating the problem by going to MAP 0050: SCSI bus problems.

64-bit PCI-X Dual Channel SCSI Adapter PTC Failure Isolation Procedure

Use the following procedures if diagnostics testing indicates a potential self-resetting thermal fuse problem. This procedure is used for SRN 2524-702 on the integrated dual-channel SCSI adapter in a 7039/651 system.

  1. Identify the adapter as the one embedded in the system board. Then, determine if the failure is on channel 0 or channel 1.
  2. The thermal fuse protects the SCSI bus from high currents due to shorts on the terminator, cable, or device. It is unlikely that the thermal fuse can be tripped by a defective adapter. A fault (short-circuit) causes an increase in resistance and temperature of the thermal fuse. The increase in temperature causes the thermal fuse to halt current flow. The thermal fuse returns to a low resistive and low temperature state when the fault is removed from the SCSI bus or when the system is turned off.

    Wait 10 seconds for the thermal fuse to reset itself and recover, then retest.

  3. If the same error persists, replace the components of the failing channel in the following order. Wait 10 seconds for the thermal fuse to reset itself between steps.
    1. Cable
    2. Device
    3. DASD backplane (if present)
    4. System board (adapter)
  4. If the failure persists, verify that the parts exchanged are in the correct channel (0 or 1). If the errors are still occurring, continue isolating the problem by going to MAP 0050: SCSI bus problems.