Hi Heinz, Since I wrote that I noticed that the drives that it fails to find the metadata on have the total number of 512-byte sectors not divisible by 63. So you could have dmraid do a search if it recognizes that case. I am pretty sure that on large 48-bit lba modern drives this will be the case where the divisor used by the Promise controller to find the last cylinder to store it's metadata is always 63, and it's a holdover from the days when the C/H/S geometry really mattered, but promise decided to keep their metadata backwards compatible with the old controllers. Thanks for your response. -James <blockquote style='border-top: 1px solid rgb(204, 204, 204); margin: 0.8ex 0pt 0pt 0pt; padding-bottom: 1ex;'> ---------[ Received Mail Content ]---------- Subject : Re: pdc type metadata location on disk Date : Mon, 28 Sep 2009 18:20:36 +0200 From : Heinz Mauelshagen <heinzm@redhat.com> To : "ATARAID (eg, Promise Fasttrak, Highpoint 370) related discussions" <ataraid-list@redhat.com> James, thanks for the heads up. We do have heuristics for a bunch of possible PDC metadata offsets but things like HPA and new offsets found keep adding to the given list. I've already thought of scanning the end of each device for any PDC signature rather than going by finite list of possible assumed ones. Probably adding such approach to a later release... Regards, Heinz On Tue, 2009-09-22 at 17:12 -0400, James Olson wrote: > Hi Heinz, > dmraid sometimes fails to find promise pdc metadata. I have analyzed > the failure and have figured out that the simple offset from the end > of the disk is not enough. The metadata is in the beginning of what > the controller thinks is the last cylinder of the disk. If you examine > the dmraid -n output, 0x22b raid.sectors: 63, in this case the place > to look for the metadata is near the end of the disk but a in sector > number divisible by 63. I have worked around this problem by adjusting > the host protected area using hdparm, but this is not an ideal > solution. > Let me know if you want more details, > -James > _______________________________________________ > Ataraid-list mailing list > Ataraid-list@redhat.com > https://www.redhat.com/mailman/listinfo/ataraid-list _______________________________________________ Ataraid-list mailing list Ataraid-list@redhat.com https://www.redhat.com/mailman/listinfo/ataraid-list </BLOCKQUOTE>