SATA thunk like noise at shutdown

[pmrdij]

hey all,

i'm hoping that someone here might have the same experience as i in my first SATA run. i just bought 4 Maxtor 80GB 8MB Cache SATA drives and when connected to their respective SATA cables i hear a thunk like noise from all of them when the power is shutdown to the system. this noise is not present at shutdown until the SATA cables are connected to the drives to the Promise S150 TX4000 RAID card i am using so i know it is not a fan or anything else.. all four emit such as i have tried them with/without the SATA cable to TX4000 connection individually.

being my first time out with SATA i phoned Maxtor's TS for the first time in 27 drives dating back to when they picked up Quantum and found nothing more than speculation from the drive having a brake for the head, power ceasing before the head is in it's final resting place, amongst other notions. i've heard unconcerning drive noises previously but this is one of those that raises the hair. both RAID Arrays i've setup bench at expected levels and run through both HdTach, Sandra and Futuremark 02/04 HDD tests without error.

anyones take as to what could spawn the noise would be helpful. even if it is more speculation will be appreciated. if need be i'll post an mp3 of the noise for further analysis as this is just weird to me. thanks!

[Ice Czar]

well a few questions first

is it a parity RAID level? (3 or 5)
if so it could well be that there isnt enough time to clear the cache writing the data to the drives before it shutsdown
and that brings the possibility you will have issues with the write hole (http://www.lostcircuits.com/hdd/hdd9/4.shtml)
and thus this could be an indication of corruption problems to come
the question would be what are they trying to write, what comes to mind would be indexing, pagefile ect

the drive should automatically park the heads when they loose power regardless
but the way they do that if in the middle of an operation could well be the cause of the sound
Head Load/Unload Technology (http://www.pcguide.com/ref/hdd/perf/qual/featuresHead-c.html)


so what is the RAID level?
what is the size of the oncard cache? (the max 256MB?)
where is the pagefile located?
do you have the pagefile set to be cleared at shutdown? (a common security tweak)
do you have whatever is on the drives indexed? (Is the indexing service enabled and are those directories set to be indexed?)

if they are in the middle of an operation, its a driver issue Id assume
as they should "clear" everything and then shutdown, with the head parked in the landing zone quitely like if the drive wasnt on the card
as opposed to loudly like a power loss "safety" head park

all this is pure guesswork ;)


possible solutions
strangely something to give the OS to do before the shutdown, might be a work around
for instance as mentioned above clearing the pagefile (which really does add a bit to the shutdown) might stretch the time required to clear the cache, turning of indexing on the array

getting a better idea of what is where would help with possible workarounds
but of course if its simply not interfacing with the OS when it comes time to shutdown, extending the period might not work anyway

[pmrdij]

well a few questions first

is it a parity RAID level? (3 or 5)
if so it could well be that there isnt enough time to clear the cache writing the data to the drives before it shutsdown
and that brings the possibility you will have issues with the write hole (http://www.lostcircuits.com/hdd/hdd9/4.shtml)
and thus this could be an indication of corruption problems to come
the question would be what are they trying to write, what comes to mind would be indexing, pagefile ect

the drive should automatically park the heads when they loose power regardless
but the way they do that if in the middle of an operation could well be the cause of the sound
Head Load/Unload Technology (http://www.pcguide.com/ref/hdd/perf/qual/featuresHead-c.html)


so what is the RAID level?
what is the size of the oncard cache? (the max 256MB?)
where is the pagefile located?
do you have the pagefile set to be cleared at shutdown? (a common security tweak)
do you have whatever is on the drives indexed? (Is the indexing service enabled and are those directories set to be indexed?)

if they are in the middle of an operation, its a driver issue Id assume
as they should "clear" everything and then shutdown, with the head parked in the landing zone quitely like if the drive wasnt on the card
as opposed to loudly like a power loss "safety" head park

all this is pure guesswork ;)


possible solutions
strangely something to give the OS to do before the shutdown, might be a work around
for instance as mentioned above clearing the pagefile (which really does add a bit to the shutdown) might stretch the time required to clear the cache, turning of indexing on the array

getting a better idea of what is where would help with possible workarounds
but of course if its simply not interfacing with the OS when it comes time to shutdown, extending the period might not work anyway
thanks for responding :). the lot of them are in 0+1 and the card itself doesn't have any option for installable cache memory on such. as to the Page File i have that on a 2GB Partition untouched by any other files and i do have the system set to clear the PF at shutdown. indexing is always shut off. thing is that this occurs with or without data on the drives (noise turns up if i power the system with all drives empty to just a DOS prompt and shut the power down.

for the most part the "safety park" notion is the only one that has come up in the last 24 hours amongst you, various cohorts, and myself that makes sense. fun times ;)

[Ice Czar]

ahh, TX not SX gotcha
(saw the last half and assumed it had a cache, Ive got an SX6000)

there are SCSI issues (which of course IDE RAID employs)
in XP and W2K w\ SP3 or later
that "might" be a contributing factor
http://www.storagereview.com/php/tiki/tiki-index.php?page=XpScsiProblems
but its unlikely to be the root cause

of course youre running the latest BIOS & Driver (1.00.0.37)