HP IO Accelerator Driver and Management Software Version 2.2.1 Release Notes

Errata 12
Errata
Errata overview
This section describes issues you might encounter when using the 2.2.1 IO Accelerator driver release.
General issues
The following issues might occur regardless of which operating system you use.
Avoid unmanaged shutdown
If the IO Accelerator experiences an unmanaged shutdown (for example, power loss or system crash), the
driver takes longer to load during the next reboot, up to several minutes. When this issue occurs, the driver
reports that it needs several minutes to rebuild management data and not to shut down or restart the system.
If you have more than one IO Accelerator, each device rebuilds the management data in parallel, rather than
serially as with previous versions.
IO Accelerator uses host memory to operate
Depending on how the IO Accelerator is configured, it uses varying amounts of host memory per 80GB of
storage capacity, as shown in the following table.
Average block
size (bytes)
RAM usage (megabytes)
8192
225
4096 (most
common)
425
2048
825
1024
1600
512 (default)
3175
Periodic latency bump
A latency bump every 30 seconds might be observed under some workloads, or a higher than expected
average latency might be observed. This issue occurs because of normal maintenance.
Under Linux, this bump can be safely changed to happen every minute instead of every 30 seconds. To set
a one minute interval, enter the following command every time you load the driver:
$ echo 60000 > /proc/fusion/fio/fiox/data/groomer/groom-oldest-lebinterval
where x is the letter of the device you are modifying.
This issue will be resolved in a future release.