Dell has warned that some Micron SSDs can slow down after 65,535 power-on hours. Is that continuous uptime, and what should admins check?
What's the deal with the Micron SSD slowdown after 65,535 hours?
Answers
It's accumulated power-on time, not time since the last reboot. 65,535 hours is about seven and a half years with the drive powered continuously, but it can also accumulate over a longer calendar period.
The number is the maximum value of an unsigned 16-bit counter. Dell describes an internal firmware limitation involving a counter used for workload log entries. The advisory is about performance degradation, so don't assume it means the drives suddenly fail at that point.
Reply to this answer
Start with an inventory of drive model numbers, firmware versions and power-on hours. Compare the actual drives against Dell's affected-model list rather than assuming every drive in a notified server is affected.
Check the advisory for firmware availability and instructions before planning an update. The discussion indicated a fix wasn't available yet for at least some affected drives, so don't assume there's already a package for every model.
_z68!marblewicket.2:
Good point about checking the drives themselves. A notification for a server doesn't necessarily tell me which installed drives match the advisory.
Reply to this answer
There have been other SSD firmware bugs tied to power-on hours, including some where drives failed to initialize after a power cycle. That's a useful reason to take the warning seriously, but it's a different failure mode from the slowdown described here.
If a system is already slow, check whether its drives and hours match the advisory before blaming the controller or assuming this is the cause. A cold power cycle was reported to temporarily relieve symptoms in one case, but that's not a firmware fix.
Reply to this answer
Add your answer
Answers are reviewed before publication.
Nervalo.5k:
So rebooting the server doesn't reset the drive's lifetime power-on hours. Checking that value on the drive is more useful than checking the server's uptime.