I wouldn’t mind seeing the mdt index be listed in the getstripe output
regardless of whether or not the file had a DoM component. That way I wouldn't
have to run "lfs getstripe --mdt" separately to get that info. But the current
getstripe output seems to closely mimic the internal data structure of the
layout, so adding the mdt index would somewhat deviate from that. I don't know
if that would be considered undesirable or not.
--Rick
On 5/29/26, 10:08 AM, "lustre-discuss on behalf of John Bauer via
lustre-discuss" wrote:
Would it be reasonable to change the output of lfs getstripe to have the mdtidx
reflected in the lmm_stripe_offset field? I have noticed that for DoM
components the value is always 0. It would make sense as the stripe_offset is
normally the OST index where striping starts for that component, and for DoM
components it could indicate the MDT where the striping starts and ends for the
component. This would add a bit of value to the lfs getstripe command without
adding extra lines. The following lfs getstripe output is for a file that
llapi_file_fget_mdtidx() reports that the mdtidx is 3.
John
% lfs getstripe b.dat
b.dat
lcm_layout_gen: 3
lcm_mirror_count: 1
lcm_entry_count: 2
lcme_id: 1
lcme_mirror_id: 0
lcme_flags: init
lcme_extent.e_start: 0
lcme_extent.e_end: 1048576
lmm_stripe_count: 0
lmm_stripe_size: 1048576
lmm_pattern: mdt
lmm_layout_gen: 0
lmm_stripe_offset: 0
lmm_pool: flash
lcme_id: 2
lcme_mirror_id: 0
lcme_flags: init
lcme_extent.e_start: 1048576
lcme_extent.e_end: EOF
lmm_stripe_count: 4
lmm_stripe_size: 1048576
lmm_pattern: raid0
lmm_layout_gen: 0
lmm_stripe_offset: 102
lmm_pool: disk
lmm_objects:
- 0: { l_ost_idx: 2, l_fid: [0x78000040e:0x10527b5:0x0] }
- 1: { l_ost_idx: 30, l_fid: [0x68000040e:0x1082064:0x0] }
- 2: { l_ost_idx: 08, l_fid: [0x48000040e:0x1052a13:0x0] }
- 3: { l_ost_idx: 18, l_fid: [0x2c000040e:0x103046f:0x0] }
_______________________________________________
lustre-discuss mailing list
[email protected]
http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org
