Asset ID:	1-72-1017987.1
Update Date:	2010-07-06
Keywords:

---

Solution Type  Problem Resolution Sure

Solution  1017987.1 :   Response to error: SepromSegment(constructor): PE at 0x----: Invalid CRC observed=xxxxxxxx expected=yyyyyyyy  


PreviouslyPublishedAs
229265

Symptoms
The following message is observed from the platform shell of a Sun Fire[TM] Server class system (F3800, F4800, F4810, F6800, V1280, N1280, E4900, E6900, E2900, N1290):

Oct 10 17:33:24 sc0 Platform.SC: [ID 417500 local0.error] /N0/SB4/P3/B1/D2: SepromSegment(constructor): PE at 0x0c3a:  Invalid CRC observed=85850b53 expected=85850b4d
Oct 10 17:33:24 sc0 Platform.SC: [ID 947597 local0.notice] Clear /N0/SB4/P3/B1/D2 invalid segment
Oct 10 17:33:42 sc0 Platform.SC: [ID 296341 local0.notice] SepromOps.clearBadSeg: /N0/SB4/P3/B1/D2 removed bad segment PE

Resolution
The error message may safely be ignored. As a board is being powered on, a PE or Power Event message is written into the FRUID of every seprom (aka seeprom - Serially Erasable E-PROM) on that board. For a system board, there is one seeprom for the board as a whole, and one on each external cache module (up to 8 of them) and one for each DIMM (up to 32 on a board).

If the system experiences a back-to-back power outage, there may be partially updated records in the FRUID data.  ScApp will detect and clean up any bad records on the next poweron event.
Should the messages persist for a particular part, the seeprom on that part may be faulty and should be replaced.
Note that in this case, the error is with the seeprom on the DIMM at /N0/SB4/P3/B1/D2.  The SDRAM on the DIMM is fine, the error is limited to a FRUID record that ScApp has written into the seeprom for that DIMM.

Relief/Workaround
The error is corrected automatically by ScApp. No action is required unless
the message persists across multiple power on events.

Additional Information
Records other than a Power Event (PE) record may also suffer from a partial write during back-to-back power outages. The most common of these, a TH or Temperature History record is also corrected automatically. Here is an example of a corrupted TH record on a Sun Fire[TM] V1280:

Tue May 18 17:36:12 c07u3cl214 lom: [ID 714981 local0.error] /N0/SB0/P0/B0/D0: SepromSegment(constructor): TH at 0x0d58:  Invalid CRC observed=08773642 expected=08a8d342
Tue May 18 17:36:13 c07u3cl214 lom: [ID 546203 local0.notice] Clear /N0/SB0/P0/B0/D0 invalid segment
Tue May 18 17:36:57 c07u3cl214 lom: [ID 636273 local0.notice] SepromOps.clearBadSeg: /N0/SB0/P0/B0/D0 removed bad segment TH

Product
Sun Fire 3800 Server
Sun Fire 4800 Server
Sun Fire 4810 Server
Sun Fire 6800 Server
Sun Fire E2900 Server
Sun Fire E4900 Server
Sun Fire E6900 Server
Sun Fire V1280 Server
Sun Netra 1290 Server
Netra 1280 Server

Internal Comments
For internal Sun use only.

The problem is further described in CR 5060437 SCapp corrupts seprom data during power cycling. It has been closed as "will not fix."
SepromSegment, Invalid CRC, seeprom, seprom, invalid segment
Previously Published As
91189

Change History
Date: 2007-10-30
User Name: 95826
Action: Approved
Comment: - changed 'serengeti' with Sun Fire[TM] as it is an internal code name and should not be used externally.
- verified metadata
- changed review date to 2008-10-29
- checked audience : contract
Publishing
Version: 3
Date: 2007-10-29
User Name: 95826
Action: Accept
Comment:
Version: 0
Date: 2007-10-28
User Name: 18410
Action: Approved
Comment: Document is a good summary of how to deal with this type of error message.
Version: 0

Attachments

This solution has no attachment