Sun Datacenter InfiniBand Switch 36
 
Sun Network QDR InfiniBand Gateway Switch
 

PLA-Support>Eng Systems>Exalogic/OVCA>Oracle Exalogic>MW: Exalogic Core
 

In this Document

Symptoms

Changes

Cause

Solution

References

Applies to:

Symptoms

Exachk script for Engineered Systems (script version 12.1.0.2.6), has a new feature which checks the "Life Expectancy" of the internal SSD-drive in the NM2 GW & 36p Switches.   Example output may show:

Unfortunately, two regression problems have been found with this new feature:

1.  The current threshold is too high (10%).   So as to align with the corresponding ASR alert, this will be re-calibrated to 2% in a future release of Exachk.  


2.  If the IB Switch firmware is lower than 2.1.x, the script may report a false-negative (since the IB Switch "showdisk" command does not exist in the earlier releases)

The internal SSD drive inside the IB Switch can only be replaced by swapping the IB Switch itself.   Replacing the IB Switch too early, represents additional inconvenience for the customer.

Changes

Changes in the environment may include that Exachk version 12.1.0.2.6 has recently been installed

Cause

Refer to Bug 22883759 for the Threshold issue.

Refer to Bug 22743695 for the missing command issue

Solution

1.   Use Exachk version 12.1.0.2.7 or higher:  The false-positive ’Life expectancy of the switch is lower than 10%’ reported in Exalogic Exachk version 12.1.0.2.6 is fixed in Exachk 12.1.0.2.7 through the bugs 22883759 / 22743695 / 21460504. If the customer runs latest Exachk 12.1.0.2.7, then they do not get false-positive alerts for 10% life expectancy, but only get the Exachk alerts, when the concerned IB switch is at 2% or lower.

OR

2.   For Exachk version 12.1.0.2.6 only:  It is essential to ask customer for the output of IB Switch "showdisk" command. If the Life expectancy indicated is greater than 2%, then reassure the customer that there is no additional risk at the current time. Even at 2%, at normal IB Switch SSD disk usage rates there may be more than 30 weeks (more than 7 months) before the indicator drops to 0%. (At 0%, the IB Switch will keep working until there is an actual SSD failure; nevertheless the Switch should be replaced at 2% or less).  Replacing the IB Switch too early represents additional cost and inconvenience for the customer.   Replacing an IB switch, often at the center of an IB Fabric, requires extensive pre-checks, may require an outage window and involves handling and movement of Fibre cables and connectors.

References

This solution has no attachment