[LU-2309] Uninteroperable conf_params not documented

• Status: Resolved
• Project: Lustre
• Affects Version/s: Lustre 2.3.0, Lustre 2.4.0, Lustre 2.1.4
• Fix Version/s: Lustre 2.4.0, Lustre 2.1.5
• Type: Bug
• Priority: Minor
• Reporter: John Hammond
• Assignee: Jian Yu
• Resolution: Fixed
• Severity: 3

Description

If a 2.3 FS has the jobid_var conf_param set then it seems (http://jira.whamcloud.com/browse/LU-1848) that 2.1/2.2 clients will be unable to mount that FS. Is this a bug or a fact of life? In either case is this documented anywhere? What other such parameters are there?

Issue Links

• is related to: LU-3452 1.8 <> 2.4 interop broken due job id ...

Comments

[Andreas Dilger]

I don't think this is a documentation issue, but rather a problem that the client should not fail to mount if it hits a configuration parameter that it doesn't understand. That needs to be fixed for 2.1.4 and 2.4.0 so that there aren't gratuitous failures.

[Peter Jones]

Yujian will look into this one

[Jian Yu]

Debug log on the 2.1.3 client showed that:

00000020:00000080:2.0:1353671368.451478:0:30937:0:(obd_config.c:1041:class_process_config()) processing cmd: cf00f
00000020:00000001:2.0:1353671368.451479:0:30937:0:(obd_config.c:927:class_set_global()) Process entered
00000020:00000001:2.0:1353671368.451480:0:30937:0:(obd_config.c:940:class_set_global()) Process leaving (rc=18446744073709551594 : -22 : ffffffffffffffea)
00000020:00000400:2.0:1353671368.451481:0:30937:0:(obd_config.c:1128:class_process_config()) Ignoring unknown param jobid_var=procname_uid
00000020:00000001:2.0:1353671368.460819:0:30937:0:(obd_config.c:1129:class_process_config()) Process leaving via out (rc=0 : 0 : 0x0)
00000020:00000001:2.0:1353671368.460820:0:30937:0:(lustre_cfg.h:212:lustre_cfg_len()) Process entered
00000020:00000001:2.0:1353671368.460821:0:30937:0:(lustre_cfg.h:218:lustre_cfg_len()) Process leaving (rc=96 : 96 : 60)
00000020:00000010:2.0:1353671368.460822:0:30937:0:(lustre_cfg.h:256:lustre_cfg_free()) kfreed 'lcfg': 96 at ffff8803301b1e40.
00000020:00000001:2.0:1353671368.460823:0:30937:0:(lustre_cfg.h:257:lustre_cfg_free()) Process leaving
00000020:00020000:2.0:1353671368.460823:0:30937:0:(obd_config.c:1488:class_config_llog_handler()) Err -22 on cfg command:
00000020:00000001:2.0:1353671368.469129:0:30937:0:(obd_config.c:1539:class_config_dump_handler()) Process entered
00000020:00000010:2.0:1353671368.469135:0:30937:0:(obd_config.c:1541:class_config_dump_handler()) kmalloced 'outstr': 256 at ffff88032c2b26c0.
00000020:00000001:2.0:1353671368.469136:0:30937:0:(lustre_cfg.h:264:lustre_cfg_sanity_check()) Process entered
00000020:00000001:2.0:1353671368.469137:0:30937:0:(lustre_cfg.h:212:lustre_cfg_len()) Process entered
00000020:00000001:2.0:1353671368.469141:0:30937:0:(lustre_cfg.h:218:lustre_cfg_len()) Process leaving (rc=96 : 96 : 60)
00000020:00000001:2.0:1353671368.469142:0:30937:0:(lustre_cfg.h:286:lustre_cfg_sanity_check()) Process leaving (rc=0 : 0 : 0)
00000020:02000400:2.0:1353671368.469145:0:30937:0:(obd_config.c:1582:class_config_dump_handler())    cmd=cf00f 0:(null)  1:sys.jobid_var=procname_uid  2:procname_uid
00000020:00000010:2.0:1353671368.476151:0:30937:0:(obd_config.c:1588:class_config_dump_handler()) kfreed 'outstr': 256 at ffff88032c2b26c0.
00000020:00000001:2.0:1353671368.476152:0:30937:0:(obd_config.c:1589:class_config_dump_handler()) Process leaving (rc=0 : 0 : 0)
00000020:00000001:2.0:1353671368.476153:0:30937:0:(obd_config.c:1491:class_config_llog_handler()) Process leaving (rc=18446744073709551594 : -22 : ffffffffffffffea)

Patch for b2_1 branch: http://review.whamcloud.com/4663

[Jian Yu]

Patch for b2_1 branch: http://review.whamcloud.com/4663

The above patch was landed on b2_1 branch. I'll create another patch to improve conf-sanity test 42 and also create the patch for master branch.

[Jian Yu]

The above patch was landed on b2_1 branch. I'll create another patch to improve conf-sanity test 42 and also create the patch for master branch.

Patch for b2_1 branch: http://review.whamcloud.com/5031.

[Jian Yu]

Patch for master branch: http://review.whamcloud.com/5037.

[Jian Yu]

The patch for Lustre master branch was landed.
The test script patch for Lustre b2_1 branch is pending review.

[Jian Yu]

The test script patch for Lustre b2_1 branch was landed.

[Niu Yawei (Inactive)]

Should this be backport to b1_8? 1.8 client doesn't understand the new param (jobid_var) neither.

[Jian Yu]

Should this be backport to b1_8? 1.8 client doesn't understand the new param (jobid_var) neither.

Patch for Lustre b1_8 branch: http://review.whamcloud.com/5972.