[LU-12894] SSK regression in 2.12.3

• Type: Bug
• Resolution: Fixed
• Priority: Major
• Fix Version/s: Lustre 2.14.0, Lustre 2.12.4
• Affects Version/s: Lustre 2.12.3
• Environment: RHEL7.6
• Severity: 3
• Assignee: Sebastien Buisson
• Reporter: Götz Waschk

Description

I have upgraded my 2.12.2 test servers to 2.12.3 and now the SSK test as described in the manual section 28.6.1 fails. I have enabled skpi for cli2mnt and cli2ost. This was working fine in 2.12.2. Now I can mount my test file system but I cannot access it:

ls: reading directory /testlustre/fs9: Permission denied

These are the client messages:

[ 3090.876407] Lustre: 13544:0:(gss_svc_upcall.c:1216:gss_init_svc_upcall()) Init channel is not opened by lsvcgssd, following request might be dropped until lsvcgssd is active
[ 3090.880312] Key type lgssc registered
[ 3091.072299] Lustre: 13552:0:(sec_gss.c:377:gss_cli_ctx_uptodate()) client refreshed ctx ffff9a7ca64c2a80 idx 0xde3c5927012269c2 (0->fs9-MDT0000_UUID), expiry 1572344698(+604790s)
[ 3091.075044] Lustre: 13552:0:(gss_svc_upcall.c:882:gss_svc_upcall_install_rvs_ctx()) create reverse svc ctx ffff9a7cae65b840 to fs9-MDT0000_UUID: idx 0x645234a1b9165cbd
[ 3092.135380] Lustre: Mounted fs9-client
[ 3092.304968] Lustre: 13557:0:(sec_gss.c:377:gss_cli_ctx_uptodate()) client refreshed ctx ffff9a7cad91f500 idx 0x1ad7691eaa6fda7b (0->fs9-OST0000_UUID), expiry 1572344699(+604790s)
[ 3092.305604] Lustre: 13560:0:(gss_svc_upcall.c:882:gss_svc_upcall_install_rvs_ctx()) create reverse svc ctx ffff9a7ca99daa40 to fs9-OST0001_UUID: idx 0x645234a1b9165cbf
[ 3092.311645] Lustre: 13557:0:(sec_gss.c:377:gss_cli_ctx_uptodate()) Skipped 1 previous similar message
[ 3095.542189] LustreError: 13587:0:(gss_bulk.c:289:gss_cli_ctx_unwrap_bulk()) failed to decrypt bulk read: 60000
[ 3101.917807] LustreError: 13593:0:(gss_bulk.c:289:gss_cli_ctx_unwrap_bulk()) failed to decrypt bulk read: 60000
[ 3107.480204] LustreError: 7196:0:(lmv_obd.c:1415:lmv_statfs()) fs9-MDT0000-mdc-ffff9a76f66c2800: can't stat MDS #0: rc = -13

 

on the MDS:

[11330.600880] Lustre: MGS: Connection restored to 44470e33-c4d6-18ae-921b-a0b0080b5f31 (at 192.168.22.13@tcp)
[11330.603719] Lustre: Skipped 3 previous similar messages
[11342.222908] Lustre: 15547:0:(sec_gss.c:2066:gss_svc_handle_init()) create svc ctx ffff9f40b548e640: user from 192.168.22.13@tcp authenticated as root
[11342.280032] Lustre: 15547:0:(sec_gss.c:370:gss_cli_ctx_uptodate()) server installed reverse ctx ffff9f43a27ccd80 idx 0x86acd6bc321ba5dd, expiry 1572344268(+604799s)
[11420.200888] Lustre: 15547:0:(sec_gss.c:2326:gss_svc_handle_destroy()) destroy svc ctx ffff9f40b548e640 idx 0xde3c5927012269c1 (0->192.168.22.13@tcp)
[11425.617420] Lustre: 12964:0:(sec_gss.c:1224:gss_cli_ctx_fini_common()) reverse sec ffff9f438c7c0000: destroy ctx ffff9f43a27ccd80
[11781.397799] Lustre: 15425:0:(sec_gss.c:2066:gss_svc_handle_init()) create svc ctx ffff9f438abe5a40: user from 192.168.22.13@tcp authenticated as root
[11781.452307] Lustre: 15425:0:(sec_gss.c:370:gss_cli_ctx_uptodate()) server installed reverse ctx ffff9f46a0a58000 idx 0x645234a1b9165cbd, expiry 1572344708(+604800s)

Comments

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch https://review.whamcloud.com/37028/
Subject: LU-12894 sec: fix checksum for skpi
Project: fs/lustre-release
Branch: b2_12
Current Patch Set:
Commit: 6fcc581a3434c3a7651515b694c675d199ea0609

[Peter Jones]

Yup - just thinking ahead

[Sebastien Buisson]

pjones I think you meant "landed for 2.14"?

Backport patch for b2_12 has just been submitted here:
https://review.whamcloud.com/37028

[Gerrit Updater]

Sebastien Buisson ([e-mail]) uploaded a new patch: https://review.whamcloud.com/37028
Subject: LU-12894 sec: fix checksum for skpi
Project: fs/lustre-release
Branch: b2_12
Current Patch Set: 1
Commit: 5d1e5ace1504455d3064be660fc4eeaa1b29515d

[Peter Jones]

Landed for 2.14

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch https://review.whamcloud.com/36604/
Subject: LU-12894 sec: fix checksum for skpi
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: dcdf060342e7d69b64171840cf9475bf65d036ea

[Gerrit Updater]

Sebastien Buisson ([e-mail]) uploaded a new patch: https://review.whamcloud.com/36604
Subject: LU-12894 sec: fix checksum for skpi
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 17f43ab90497bd7ec521d89029688541cb992ef3

[Sebastien Buisson]

Hi,

I can reproduce this issue. I think I have identified the problem, and I might be able to push a fix soon.

[Peter Jones]

Sébastien

Can you please advise

Thanks

Peter