ID: 2898
Title: smart: Using normalized value for determining reallocated events state
Component: Checks & Agents
Level: 1
Class: Bug Fix
Version: 1.2.7i4
The smart check often reported errors for disks which had just a small number of
reallocated events but where the normalized state value (aggregated by hardware
manufacturer logic) did not report an issue.
This was caused by the logic of the check which stores values the agent reports
during service discovery and continously checks the difference between these values
and the current values reported by the disk. In this progress the "raw values"
of
the smartctl output were used for all attributes.
But this seemed - at least for the "reallocated events count" - not a really
good
approach because the raw values may slightly change during regular operation. But
small changes of the values does not necessarly mean a predicted failure. So we
decided to change the logic only for the "reallocated events" now.
Maybe we need to change it for the other values in the future. But for the moment
only the single value interpretation was changed.