I believe the default PERC setting is to leave the disk cache at the
default setting. However, can manually enable/disable the disk cache on
a per VD basis to see if that has any positive or negative effect (this
change can be made via perccli or the iDrac). For data integrity
reasons, most folks recommend disabling the disk's cache to ensure data
is committed promptly.
--Blake
On 5/18/2021 1:19 PM, Paul Raines wrote:
Thanks, maybe the cache is the reason. I am still suprised by the
difference if that is it.
The Seagate RAID is progressing at about 0.25% per hour while the
WD RAID is progressing about 2% per hour. So 8x faster.
-- Paul Raines (http://help.nmr.mgh.harvard.edu)
On Tue, 18 May 2021 1:53pm, Blake Hudson wrote:
External Email - Use Caution The WD drives are a family that
comes in either 18TB or 16TB sizes while the Seagate model comes in
16TB-10TB sizes, this would indicate that the WD drives have higher
density or might be based on newer technology. The WD drives have a
512MB cache while the Seagate drives have 256MB cache; The increased
cache may provide additional performance in a wider variety of
applications.
As far as Dell is concerned, a 16TB 7.2k SAS drive is a commodity
part where one model/brand is as good as another, even if the
replacement part is a model based on technology several years older
or newer than the original.
https://secure-web.cisco.com/1pHUYOg2xL6iDLuEfdSrcWvN_mBPTKnJPQ_IoHNDbouMN4VhWXX2FKDn6p-siRkpZnYE4umdmynorozZHhBR4XkPcY4s3oKu_GLqBfK8ou7ElLuVa-NTYq0x7vBGM5exQnVyI52fTdp0sRoic5hD_6Hnm59YySdDfBxmsETMpDwjnTFOKrehyJN4af-PT6_fyhQ7oGPXmc7-0dF88g_2-upnYRX31dBQxK6HLtDWaVV8rVUycwQJpXJUF6BrCqSBt2wvpA6K_4tBoEHXilWeffg/https%3A%2F%2Fwww.seagate.com%2Ffiles%2Fwww-content%2Fdatasheets%2Fpdfs%2Fexos-x16-DS2011-3-2008US-en_US.pdf
https://secure-web.cisco.com/1YhZeVbxD3JZU-Gss1FRY5CuMSxy6G3Llst8bHzTk6ujgVe20gzkXNSu-wnwor6C1kMCpvqRdKak2D2rpx3misb5fqkwRd4jf_ZnsTlSNB9qmJAFMNuc4pwMoQ1iif1LYTLxtSdrYZDpREKDUWIPui-etDgVUib7-KujOLMQTMfb_HnJut8uQRwnf2OXIk05AwXgtbdoe2sJgkbttMwS9Y5jphGLDbh4O3IRvVo57X030MC1R-9iM8-hIfiAIJ3OlMmX4WxDVOujn00s9pYskfQ/https%3A%2F%2Fdocuments.westerndigital.com%2Fcontent%2Fdam%2Fdoc-library%2Fen_us%2Fassets%2Fpublic%2Fwestern-digital%2Fproduct%2Fdata-center-drives%2Fultrastar-dc-hc500-series%2Fproduct-manual-ultrastar-dc-hc550-sas-oem-spec.pdf
--Blake
On 5/18/2021 11:57 AM, Paul Raines wrote:
I did firmware updates of everything in the lifecycle controller
as the first thing I did when I booted before making the RAIDs
so all firmware levels should be the same.
The disks are different brands it appears. Not sure why that
happened as we ordered these at the same time.
The first, slower box has ST16000NM010G 16TB drives and
the second, faster box has WUH721816AL5200 16TB drives.
A suprising difference to have different brands of disks but I am also
suprised it makes such a big difference in the init rates if
that really is the cause.
I guess I will open a ticket will Dell support.
-- Paul Raines
(http://secure-web.cisco.com/1utkIp4YPXrFeKb7D1ttrCwJqycVgCZJXU69Wev8U-OwUMPp-OC3gvVVDncwbXWXHikHT-fq_lFeleAXy9MlvWBl4wcST5SAZaSHoB-KniG74INsRrJWJvkUXmAtq5KmAQfIHfDBVY2sNSC1Sds08I5-QCHOmGPa_zHs7ETqUFWIJ_hNuTDYKxCD2_XBv4BpJ29BTaBbMol9_BuoVGFNE18oqmNUAyviVhEjJVU5pduEkn3dnJTMw_N9EA7-1fL5TTCGCNaP6UqUDp2Ma1wae7w/http%3A%2F%2Fhelp.nmr.mgh.harvard.edu)
On Tue, 18 May 2021 11:42am, Blake Hudson wrote:
External Email - Use Caution One explanation might be that the
disks are not identical. Dell will sometimes ship different model
drives
depending on availability, or the drives could have come with a
different
firmware version installed. You might check with perccli to see if
this
is a possibility.
Also, some manufacturers may re-use the same model number to
indicate a
drive of X capacity for X usage, but the internals for the drive
could
change from year to year (WD is known for this). Something to
check if
the drives report a subtly different model number, different date of
manufacture/location of manufacture, or have different style
labels or
fonts from one server to another.
--Blake
On 5/18/2021 9:20 AM, Paul Raines wrote:
I recently bought two identical PowerEdge T640 with internal
PERC H730P
Adapter and eighteen 16TB disks. I created a 17 disk RAID6 with
1 hot
spare on both.
On the 1st server I created the RAID Sunday (May 16 16:20:13)
and on the 2nd server I creatd it Monday (May 17 11:23:45)
As of right now the 1st server RAID reports 9% complete while
the 2nd server RAID reports 50% complete. Which is crazy since
the 1st has been running almost half a day longer!
I checked the BGI rate on both which are both the default 30%
There is no other activity on either server going on as I am
waiting for the init to finish.
I can find no errors or other clues in the logs I know to check.
All other aspects of the RAIDs are good (all disks report good)
A comparison of 'perccli64 /c0 show all' between the two shows
no differences in any settings.
Simple random-write 'fio' tests on both show slightly slower
on the first server compared to the second but only about 10-15%
Anyone have any clues as to why the huge slowness on the
initialization on the first server?
_______________________________________________
Linux-PowerEdge mailing list
Linux-PowerEdge@dell.com
https://lists.us.dell.com/mailman/listinfo/linux-poweredge
