Sun Storage 6920 System
 

GCS>Sun Microsystems>Sun Alert>Criteria Category>Availability
 
GCS>Sun Microsystems>Sun Alert>Release Phase>Resolved
 

Under certain conditions, power cycling an SE6920 rack may render the array inaccessible, causing loss of access to volumes and data, loss of management access to volumes, and/or incorrect volume status.

This issue can occur on the following platform:

• Sun StorEdge 6920 array with DSP firmware patch before 118396-52

Notes:

1. This issue can occur with all current DSP firmware for the Sun StorEdge 6920 with DSP firmware patch before 118396-52.
2. Firmware versions are not Solaris dependent.

This issue can occur when the DSP is rebooted before or about the same time as the 6920 array. This condition may be triggered by loss of power (power black-out) to the SE6920 or partial or complete power shutdown initiated from SE6920 configuration services.

Should the described issue occur, SE6920 volumes will be in an incomplete or missing state. From the Sun StorEdge 6920 Configuration Service GUI (under the "Logical Storage Volumes" tab) the state of some or all volumes will be missing or incomplete. Under this same tab, the condition may be one of the following (depending on the configuration):

• Broken - The volume is not currently operational
• Degraded - The volume is degraded, and one or more input or output data paths is not operating properly; however, the redundant failover paths are still intact
• Incomplete - The volume configuration is incomplete, most likely due to previous failures, possibly in its configuration
• Lost Communications - The volume is offline, most likely due to a loss of communications with part or all of the storage that makes up that volume

Directions for collecting a "solution extract" can be found in the Help section of the Configuration Services GUI. To get to the the help section:

1. Log in to the web console and select the "Help" button on the upper right hand corner of the GUI.
2. In the new window that pops up, select the "Search" tab on the upper left hand side.
3. In the search window that appears type in "solution extract" and select the search button. A list of options will be displayed in GUI including "Generating a Solution Extract."
4. Select the "Generating a Solution Extract" link for directions on how to collect the extract.

From the solution extract output, the show volumes -detailed file can have entries similar to the following:

    vol/ORGA_archive   SAN    N/A    N/A MISSING    N/A
    N/A    N/A    N/A    ORGA    60:00:15:D0:00:03:5D:00:00:00:00:00:00:00:11:10
    vol/ORGA_backup    SAN    N/A    N/A MISSING    N/A
    N/A    N/A    N/A    ORGA    60:00:15:D0:00:03:5D:00:00:00:00:00:00:00:11:18

Also, vdisks may have path information missing or no active path listed and the state will be "Free" instead of "In-Use", as in the following example (from the extractor):

    Vendor Name    Product ID Raw/Initialized      Size
    -----------    --------------------------      ----
       SUN                   T4                136.52 GB/136.40 GB
    Proc      Path         Flags      State        Port WWN
    ----      ----         -----      -----        --------
    3-3      disk/3/6/2/0            Blocked       20:03:00:03:BA:68:F1:15
    4-3      disk/4/6/3/0    P       Initialized   20:03:00:03:BA:CC:BD:09
    Slice    State    Slice Size   Free Space Total/Max
    -----    -----    ----------   --------------------
      0       Free     136.40            N/A

To work around the described issue, a reboot of the DSP will temporarily resolve this issue until a permanent fix can be applied.

Note: If the SE6920 has SSRR enabled, a remote services engineer can log in to the array and reboot the DSP (Log in and password information is typically only available to SUN personnel).

To reboot the DSP, log in to the NTC through the serial connecton, and connect to the DSP from the NTC by using the following commands:

    ntc0: connect local port_3
    Local protocol emulation 1.0  - Local Switch: <\1B>
    dsp00#

Once connected to the DSP, execute a "reboot now". You will be asked if you want to overwrite the configuration file; accept the default response of yes [Y]:

    dsp00# reboot now
    Are you sure you want to reboot the system[N]? y
    OK to overwrite configuration file[Y]? y

This issue is addressed on the following platform:

• Sun StorEdge 6920 with DSP firmware patch 118396-52 or later

30-Nov-2006:

• Updated Contributing Factors and Resolution sections

References

See also BugID 6214084.

See also InfoDoc 83343 at http://sunsolve.central.sun.com/search/document.do?assetkey=1-9-83343-1

Optional Workaround for qualified personnel only:

If a planned power reset is to take place then before the array is power back up, unplug the two power cables from the DSP, wait about 7 minutes / long enough for every component apart from the DSP to come up on it's own, and plug the power-cables back into the DSP.

This process is to avoid the problem by not allowing the DSP to come online before the arrays (6020's) are up and running. The DSP doesn't have a power switch so the only 2 ways to power off the DSP is to trip the circuit breakers on the rack (which turns off all the components) or to pull the power cords from the DSP. By pulling the power cords on the DSP and powering on the rest of the rack, the arrays in the rack (6020's) will come up and be available before the DSP. Once the arrays are up (after seven minutes) the DSP can be brought online by plugging in the power and all the components (volumes) should be discovered.

References

This solution has no attachment