After a domain crash on the E25k, normally the domain will restart. Such as the case earlier this week with the oradev1 domain on the DC1 E25k. Problem determination uncovered one CPU experienced a corrupted ecache location which when detected immediately causes a system panic crash. Diagnostics for root cause normally are performed in the hours well past the system recovers. In this case the system had been restarted and running 24 hours by the time of pinpointing root cause. Protocol is to replace such failing components rather than risk a recurrence.
↧