[LU-4239] lfs fid2path ioctl err -75: Value too large for defined data type (75)

• Type: Bug
• Resolution: Fixed
• Priority: Minor
• Fix Version/s: Lustre 2.8.0
• Affects Version/s: Lustre 2.6.0
• Labels: cdwgbl patch
• Environment: lustre-client-2.4.0-2.6.32_358.11.1.el6.x86_64_ga049816.x86_64
• Severity: 3
• Assignee: Alex Zhuravlev
• Reporter: Gabriele Paciucci (Inactive)

Description

I have created 100 nested directories. No problem until 99.

[root@st01-cli2 ~]# lfs fid2path /mnt/pippo [0x200000400:0xcc:0x0]
/mnt/pippo/1/2/3/4/5/6/7/8/9/10/11/12/13/14/15/16/17/18/19/20/21/22/23/24/25/26/27/28/29/30/31/32/33/34/35/36/37/38/39/40/41/42/43/44/45/46/47/48/49/50/51/52/53/54/55/56/57/58/59/60/61/62/63/64/65/66/67/68/69/70/71/72/73/74/75/76/77/78/79/80/81/82/83/84/85/86/87/88/89/90/91/92/93/94/95/96/97/98/99

[root@st01-cli2 ~]# lfs fid2path /mnt/pippo [0x200000400:0xcd:0x0]
ioctl err -75: Value too large for defined data type (75)
fid2path error: Value too large for defined data type

Issue Links

• is related to: LU-6384 sanity test_154g: llapi_fid2path failed for fid 0x2000013a1:0x3b:0x0: Invalid argument
• is related to: LU-6194 Interop 2.5.3<->2.7 sanity test_162c: Value too large for defined data type

Comments

[Gerrit Updater]

"Oleg Drokin <[e-mail]>" merged in patch https://review.whamcloud.com/c/fs/lustre-release/+/58047/
Subject: LU-4239 tests: add missing call to test
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 8cd8f3f64d2034ea600e5ca83bf13faff0e623d9

[Gerrit Updater]

"Robert Read <[e-mail]>" uploaded a new patch: https://review.whamcloud.com/c/fs/lustre-release/+/58047
Subject: LU-4239 tests: add missing call to test
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: eda017c760c4d6c5ac1cfe1a02c2bb25dc828292

[Peter Jones]

Landed for 2.8

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch http://review.whamcloud.com/12545/
Subject: LU-4239 tests: test FID related APIs
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 2b10c0bde269e1d45ae9483a492ce171f05831a1

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch http://review.whamcloud.com/10717/
Subject: LU-4239 mdt: Optimize fid2path
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 825570b0b03e0fd70aba5400dedca4e6d94d5141

[Gerrit Updater]

Patrick Farrell ([e-mail]) uploaded a new patch: http://review.whamcloud.com/10717
Subject: LU-4239 mdt: Optimize fid2path
Project: fs/lustre-release
Branch: master
Current Patch Set: 15
Commit: 78034fb07174a1fe9e22fa5ea54a58d6007af0b3

[Frank Zago (Inactive)]

Tests for this patch: http://review.whamcloud.com/12545

[James A Simmons]

Can you add a test so Maloo can verify this patch.

[Patrick Farrell (Inactive)]

Patch is here:
http://review.whamcloud.com/#/c/10717/

PS1 increases the array size, PS2 is a rewrite to no longer use the array.

Testing here:
centclient01 is running a recent version of master,
centss10 is also running a recent version of master.
LU-4239 is present, as seen in the "Value too large" error.

centss09 is running a recent version of master+my patch set 2.

[root@centclient01 centss09]# cd /mnt/centss10
[root@centclient01 centss10]# lfs path2fid 1/2/3/4/5/6/7/8/9/10/11/12/13/14/15/16/17/18/19/20/21/22/23/24/25/26/27/28/29/30/31/32/33/34/35/36/37/38/39/40/41/42/43/44/45/46/47/48/49/50/51/52/53/54/55/56/57/58/59/60/61/62/63/64/65/66/67/68/69/70/71/72/73/74/75/76/77/78/79/80/81/82/83/84/85/86/87/88/89/90/91/92/93/94/95/96/97/98/99/100/101/102/103/104/105/106/107/108/109/110/
[0x200000400:0x6f:0x0]
[root@centclient01 centss10]# lfs fid2path /mnt/centss10/ [0x200000400:0x6f:0x0]
ioctl err -75: Value too large for defined data type (75)
fid2path: error on FID [0x200000400:0x6f:0x0]: Value too large for defined data type
[root@centclient01 centss10]# cd ../centss09
[root@centclient01 centss09]# lfs path2fid 1/2/3/4/5/6/7/8/9/10/11/12/13/14/15/16/17/18/19/20/21/22/23/24/25/26/27/28/29/30/31/32/33/34/35/36/37/38/39/40/41/42/43/44/45/46/47/48/49/50/51/52/53/54/55/56/57/58/59/60/61/62/63/64/65/66/67/68/69/70/71/72/73/74/75/76/77/78/79/80/81/82/83/84/85/86/87/88/89/90/91/92/93/94/95/96/97/98/99/100/101/102/103/104/105/106/107/108/109/110/
[0x200000400:0x74:0x0]
[root@centclient01 centss09]# lfs fid2path /mnt/centss09 [0x200000400:0x74:0x0]
/mnt/centss09/1/2/3/4/5/6/7/8/9/10/11/12/13/14/15/16/17/18/19/20/21/22/23/24/25/26/27/28/29/30/31/32/33/34/35/36/37/38/39/40/41/42/43/44/45/46/47/48/49/50/51/52/53/54/55/56/57/58/59/60/61/62/63/64/65/66/67/68/69/70/71/72/73/74/75/76/77/78/79/80/81/82/83/84/85/86/87/88/89/90/91/92/93/94/95/96/97/98/99/100/101/102/103/104/105/106/107/108/109/110
[root@centclient01 centss09]#

[Patrick Farrell (Inactive)]

Expanding a little on the possible concern about memory usage.

If this is a commonly used piece of code, we've gone from a struct that probably fits in one page to one that doesn't fit in 8. That's potentially quite bad, if it's in a fairly hot path on the MDS. I'm hoping some Intel engineers can comment on that as a possible problem, and if it is one, suggest other solutions.

For example - Do we need to record FIDs in this manner when looking up paths? I see it being done, but I don't actually see the data stored in pli_fids being used. The array is accessed, but only the most recent values in it. It looks like it would be simple enough to rewrite the loop in mdt_path_current to not use the array.