希捷hdd部分smart属性分析

Attribute ID 1: Raw Error Rate
Raw Usage
Raw [3 – 0] = Number of sector reads
Raw [6 - 4] = Number of read errors.
Normalized
Raw Error Rate = 10 * log10(NumberOfSectorsTransferredToOrFromHost * 512 * 8 / (Number of sectors requiring retries))
可以看到smart 1的原始值并不是字面意思“原始错误率”，原始值由7个字节组成，低四字节为read的扇区数量，高三字节是read出错的数量，因此高三位才可能代表磁盘的故障情况。
标准化后的值由NumberOfSectorsTransferredToOrFromHost 和Number of sectors requiring retries（Attribute ID 5）计算得到，其值代表了发生sector需要requiring retries的频率，另外只有在NumberOfSectorsTransferredToOrFromHost * 512 * 8（也叫做NumberOfBitsTransferredToOrFromHost ）处于10e10-10e12范围内时，Normalized Raw Error Rate才会被计算，初始值为100。
normalized的值变化趋势如下

如上图所示，在1时刻，其初始值为100，在1时刻的时候其NumberOfBitsTransferredToOrFromHost达到了10e10，Normalized Raw Error Rate开始被计算，因为在刚开始Number of sectors requiring retries的存在，使Normalized Raw Error Rate值突然变小，从2时刻开始，NumberOfBitsTransferredToOrFromHost慢慢变大，Normalized Raw Error Rate值也开始变大，从6时刻开始，NumberOfBitsTransferredToOrFromHost和Number of sectors requiring retries都开始变大，Normalized Raw Error Rate的值随之变小，从13时刻开始Number of sectors requiring retries不再变化，NumberOfBitsTransferredToOrFromHost开始变大，Normalized Raw Error Rate值随之变大，到22时刻NumberOfBitsTransferredToOrFromHost的值达到了10e12，于是NumberOfBitsTransferredToOrFromHost随之清零.

Attribute ID 5: Retired Sectors Count
原始值
Raw [1 – 0] = Current Retired Sector Count
Raw [3 - 2] = Current Retired Sector Count since SMART was last reset
可以看到原始值由两部分组成，低2字节代表了当前Retired Sector的数量，高两位还不清楚是什么意思
Normalized Retired Sectors Count = 100 - (100 * NumberOfRetiredSectors /
(MinimumNumberOfSparesAvailable))
normalized的值为原始值低两位和MinimumNumberOfSparesAvailable计算得到，MinimumNumberOfSparesAvailable根据厂家而不同

Attribute ID 7: Seek Error Rate
Raw Usage
Raw [3 – 0] = Number of seeks
Raw [5 – 4] = Number of seek errors
可以看到原始值由6字节组成组成，只有高两字节才能代表磁盘的故障情况
Normalized
Seek Error Rate = 10 * log10(SeekCount / SeekErrors)
normalized的值由SeekCount（原始值中的低四位）和SeekErrors（原始值中的高两位），而且normalized的只在在SeekCount处于10e6-10e9区间时才会被计算，当SeekCount大于10e9时，其值会被清零。

上图中的初始值位250，我的磁盘的初始值为90，在接近5时刻，SeekCount的值达到了10e6，则Normalized Seek Error Rate开始更新，因此Normalized Seek Error Rate的值会大幅度下降，在5-15时刻，SeekCount的值慢慢变大，Normalized Seek Error Rate的值也随之变大，在接近15时刻时，SeekErrors和SeekCount都在变大，Normalized Seek Error Rate开始变小，从20时刻开始SeekErrors停止增长，SeekCount继续变大，Normalized Seek Error Rate也随之变大，在接近35时刻，SeekCount达到10e9，Normalized Seek Error Rate随之被置为初始值，SeekCount也被清零。

Attribute ID 188: Command Timeout Count
Raw Usage
Raw [1 – 0] = Total # of command timeouts, with Max hold of FFFFh
Raw [3 – 2] = Total # of commands with > 5 second completion, including those > 7.5 seconds
Raw [5 – 4] = Total # of commands with > 7.5 second completion
可以看到原始值由6个字节组成，如果该值过大，可以只看低两字节。

Normalized Command Timeout Count = 100 – Command Timeout Count

Attribute ID 190: Airflow Temperature
Normalized Airflow Temperature = 100 – Degrees Celsius
Raw Usage
Raw [1 - 0] = Current Temperature in degrees Celsius
Raw [2] = Lowest temperature recorded during this power cycle
Raw [3] = Highest temperature recorded during this power cycle
Raw [5 – 4] = Number of Times attribute below Threshold
(i.e. “High Temperature” conditions). Max Hold FFFF

Attribute ID 194: Temperature
Normalized Temperature = Raw Temperature in Celsius + Temperature Variation, which give a
good correlation to the HDA Temperature.
Raw Usage
Raw [1 – 0] = Current Temp
Raw [5 – 4] = Worst case Coldest Temperature

Attribute ID 195: ECC On the Fly Count
Raw Usage
Raw [3 – 0] = Number of Sector reads
Raw [6 - 4] = Number of ECC OTF errors .
Normalized ECC On the Fly Count = 10 * log10(NumberOfSectorsTransferredToOrFromHost *
512 * 8 / (Number of On-The-Fly ECC corrected sectors ))
NumberOfSectorsTransferredToOrFromHost * 512 * 8 = NumberOfBitsTransferredToOrFromHost
只有当NumberOfBitsTransferredToOrFromHost 处于10e10-10e12时，该Normalized ECC On the Fly Count才会被计算

Attribute ID 197: Pending-Sparing Count
Normalized Pending-Sparing Count = 100 – (100*PendingSectorCount/DefectTableEntries)
Raw Usage
Raw [1 – 0] = Current Pending Spare Count
Raw [3 - 2] = Current Pending Spare Count since SMART was last reset