Hello,
sorry if the question was already asked. The search didn't find anything.
How does HD Sentinel query disks? Does it only query the SMART data every <detection_frequency> minutes?
In other words, will HD Sentinel wake a disk from standby by querying it? Assuming the disk supports the feature of saving SMART data prior to going to sleep.
Thanks
Are disks woken up by querying them?
-
hdsentinel
- Site Admin
- Posts: 3115
- Joined: 2008.07.27. 17:00
- Location: Hungary
- Contact:
Re: Are disks woken up by querying them?
Yes, this topic already discussed in different topics 
I wonder why the search did not find anything, I just did a quick search for the word wake and found results.
Generally when the detection should happen (based on the detection frequency setting on the Configuration -> Advanced page) Hard Disk Sentinel checks the current power state of the device first - and if the hard disk seems active (not sleeping) then yes, it detects the self-monitoring data, including the temperature, error-counters, statistics and so.
It designed to do not wake up sleeping drives just to detect their status.
The Information page shows the actual power state: Active or Sleeping since ... (the date/time when Hard Disk Sentinel last detected active state)
The problem is that many hard disks (and even some controllers/chipsets, regardless of the actual hard disk) do not provide accurate power state information:
- the drive may always seems active (even if it is sleeping)
- the drive may always seems sleeping (even if it is active)
and then we have no proper power state information. Then the affected disk may wake up
I wonder why the search did not find anything, I just did a quick search for the word wake and found results.Generally when the detection should happen (based on the detection frequency setting on the Configuration -> Advanced page) Hard Disk Sentinel checks the current power state of the device first - and if the hard disk seems active (not sleeping) then yes, it detects the self-monitoring data, including the temperature, error-counters, statistics and so.
It designed to do not wake up sleeping drives just to detect their status.
The Information page shows the actual power state: Active or Sleeping since ... (the date/time when Hard Disk Sentinel last detected active state)
The problem is that many hard disks (and even some controllers/chipsets, regardless of the actual hard disk) do not provide accurate power state information:
- the drive may always seems active (even if it is sleeping)
- the drive may always seems sleeping (even if it is active)
and then we have no proper power state information. Then the affected disk may wake up
Re: Are disks woken up by querying them?
If the drive supports "(0x0003) Saves SMART data before entering power-saving mode", it will not be woken up by querying of SMART attributes, e.g. 'smartctl -a' does not wake such drives. Would it make sense for HDS to skip the power state check for such drives and simply do the probing?hdsentinel wrote:Yes, this topic already discussed in different topics
Generally when the detection should happen (based on the detection frequency setting on the Configuration -> Advanced page) Hard Disk Sentinel checks the current power state of the device first - and if the hard disk seems active (not sleeping) then yes, it detects the self-monitoring data, including the temperature, error-counters, statistics and so.
It designed to do not wake up sleeping drives just to detect their status.
The Information page shows the actual power state: Active or Sleeping since ... (the date/time when Hard Disk Sentinel last detected active state)
The problem is that many hard disks (and even some controllers/chipsets, regardless of the actual hard disk) do not provide accurate power state information:
- the drive may always seems active (even if it is sleeping)
- the drive may always seems sleeping (even if it is active)
and then we have no proper power state information. Then the affected disk may wake up
For those that don't support this feature, a power state check of course make sense.
-
hdsentinel
- Site Admin
- Posts: 3115
- Joined: 2008.07.27. 17:00
- Location: Hungary
- Contact:
Re: Are disks woken up by querying them?
> If the drive supports "(0x0003) Saves SMART data before entering power-saving mode", it will not be woken up by querying of SMART attributes
I'm afraid no, this is completely different thing.
Generally all modern hard disk drives support and work with this feature, except some special models and/or if this is turned off.
This function controls if the hard disk drive should automatically save the current self-monitoring SMART data BEFORE entering to power save (sleep) mode. This is widely enabled: as we'd expect the drive should record any possible problems or just updates in statistical attributes (for example power on time, amount of data read/written and so).
But this does not affect how the detection of self-monitoring (reading the SMART data) is processed AFTER entering to power-save mode. Detection of SMART information (like any read command) causes wake up of the hard disk.
> , e.g. 'smartctl -a' does not wake such drives.
No, the drive wakes up.
Even if the above would mean that SMART-detection alone does not wake up the drive, smartctl -a performs not only that command, but also performs IDENTIFY DEVICE command too, to first identify the drive and its features (eg. to detect if S.M.A.R.T. is supported at all).
And this command (in addition to the SMART detection) causes the drive to wake up.
Hard Disk Sentinel only performs identification on startup (and on possible hardware changes to react to the changed configuration) but on periodic deection, yes, it detects SMART information only, exactly to minimise the data communication cased by the status detection, minimise the chance of wake up (if the drive can really provide the S.M.A.R.T. information without wake up).
Generally if a drive can provide SMART information without wake up, then of course it remains sleeping when monitored by Hard Disk Sentinel.
But this depends on many factors and even if sounds surprising, usually completely independent from the hard disk itself, but more related to
- disk controller and its driver (where the hard disk connected)
- possible USB adapter / enclosure / docking station
- possible RAID controller / backplane
as these all affect how / which commands are transferred to the hard disk itself, how the commands may be "coverted" or even the response cached as some USB adapters may cache S.M.A.R.T. information and provide that on periodic detection. This causes that the hard disk remains sleep mode as if the USB enclosure (which manages the power state of the drive) puts the drive in sleep mode, then the cached S.M.A.R.T. information provided.
This internal power management function is completely independent from the Winows power management setting and the type/functionality/features of the hard disk too.
> Would it make sense for HDS to skip the power state check for such drives and simply do the probing?
No, I see no point in changing or making things worse than current.
Hard Disk Sentinel performs an additional check before accessing the hard disk, just to be sure, exactly to prevent drives waking up if possible, if the controller and driver provides the power state correctly. As I know, smartctl performs similar but not sure.
And if the drive (or the enclosure, the controller, the driver, etc. ) can provide the self-monitoring information without waking up the drive, the drive remains in sleep mode.
> For those that don't support this feature, a power state check of course make sense.
Generally all drives support this feature and enabled - but it has no effect about how S.M.A.R.T. data reported when the drive already entered sleep mode.
I'm afraid no, this is completely different thing.
Generally all modern hard disk drives support and work with this feature, except some special models and/or if this is turned off.
This function controls if the hard disk drive should automatically save the current self-monitoring SMART data BEFORE entering to power save (sleep) mode. This is widely enabled: as we'd expect the drive should record any possible problems or just updates in statistical attributes (for example power on time, amount of data read/written and so).
But this does not affect how the detection of self-monitoring (reading the SMART data) is processed AFTER entering to power-save mode. Detection of SMART information (like any read command) causes wake up of the hard disk.
> , e.g. 'smartctl -a' does not wake such drives.
No, the drive wakes up.
Even if the above would mean that SMART-detection alone does not wake up the drive, smartctl -a performs not only that command, but also performs IDENTIFY DEVICE command too, to first identify the drive and its features (eg. to detect if S.M.A.R.T. is supported at all).
And this command (in addition to the SMART detection) causes the drive to wake up.
Hard Disk Sentinel only performs identification on startup (and on possible hardware changes to react to the changed configuration) but on periodic deection, yes, it detects SMART information only, exactly to minimise the data communication cased by the status detection, minimise the chance of wake up (if the drive can really provide the S.M.A.R.T. information without wake up).
Generally if a drive can provide SMART information without wake up, then of course it remains sleeping when monitored by Hard Disk Sentinel.
But this depends on many factors and even if sounds surprising, usually completely independent from the hard disk itself, but more related to
- disk controller and its driver (where the hard disk connected)
- possible USB adapter / enclosure / docking station
- possible RAID controller / backplane
as these all affect how / which commands are transferred to the hard disk itself, how the commands may be "coverted" or even the response cached as some USB adapters may cache S.M.A.R.T. information and provide that on periodic detection. This causes that the hard disk remains sleep mode as if the USB enclosure (which manages the power state of the drive) puts the drive in sleep mode, then the cached S.M.A.R.T. information provided.
This internal power management function is completely independent from the Winows power management setting and the type/functionality/features of the hard disk too.
> Would it make sense for HDS to skip the power state check for such drives and simply do the probing?
No, I see no point in changing or making things worse than current.
Hard Disk Sentinel performs an additional check before accessing the hard disk, just to be sure, exactly to prevent drives waking up if possible, if the controller and driver provides the power state correctly. As I know, smartctl performs similar but not sure.
And if the drive (or the enclosure, the controller, the driver, etc. ) can provide the self-monitoring information without waking up the drive, the drive remains in sleep mode.
> For those that don't support this feature, a power state check of course make sense.
Generally all drives support this feature and enabled - but it has no effect about how S.M.A.R.T. data reported when the drive already entered sleep mode.