[LU-3199] Interoperability problem between Lustre 2.1.3 clients and lustre 2.3.63 servers

• Status: Resolved
• Project: Lustre
• Affects Version/s: Lustre 2.4.0, Lustre 2.1.3
• Type: Bug
• Priority: Major
• Reporter: Diego Moreno (Inactive)
• Assignee: WC Triage
• Resolution: Not a Bug
• Severity: 3

Description

I'm testing the 2.3.63 tag with a 2.1.3 client I can't mount this client.

All servers are installed in 2.3.63.

My Lustre client is in 2.1.3.

Once my MDS/MGS is started and my OSS too I cannot mount the Lustre filesystem.

On MDS/MGS:

[root@falcao2 ~]# lctl dl
0 UP osd-ldiskfs MGS-osd MGS-osd_UUID 5
1 UP mgs MGS MGS 7
2 UP mgc MGC10.5.0.2@tcp 2d7d64f3-854c-60f8-ba9f-b04827fb7012 5
3 UP osd-ldiskfs falcaofs-MDT0000-osd falcaofs-MDT0000-osd_UUID 9
4 UP mds MDS MDS_uuid 3
5 UP lod falcaofs-MDT0000-mdtlov falcaofs-MDT0000-mdtlov_UUID 4
6 UP mdt falcaofs-MDT0000 falcaofs-MDT0000_UUID 9
7 UP mdd falcaofs-MDD0000 falcaofs-MDD0000_UUID 4
8 UP qmt falcaofs-QMT0000 falcaofs-QMT0000_UUID 4
9 UP osp falcaofs-OST0001-osc-MDT0000 falcaofs-MDT0000-mdtlov_UUID 5
10 UP osp falcaofs-OST0002-osc-MDT0000 falcaofs-MDT0000-mdtlov_UUID 5
11 UP lwp falcaofs-MDT0000-lwp-MDT0000 falcaofs-MDT0000-lwp-MDT0000_UUID 5

On OSS:

[root@falcao3 ~]# lctl dl
0 UP osd-ldiskfs falcaofs-OST0002-osd falcaofs-OST0002-osd_UUID 5
1 UP mgc MGC10.5.0.2@tcp 15818cd1-7577-8e13-aa40-601eca445800 5
2 UP ost OSS OSS_uuid 3
3 UP obdfilter falcaofs-OST0002 falcaofs-OST0002_UUID 5
4 UP osd-ldiskfs falcaofs-OST0001-osd falcaofs-OST0001-osd_UUID 5
5 UP obdfilter falcaofs-OST0001 falcaofs-OST0001_UUID 5
6 UP lwp falcaofs-MDT0000-lwp-OST0001 falcaofs-MDT0000-lwp-OST0001_UUID 5
7 UP lwp falcaofs-MDT0000-lwp-OST0002 falcaofs-MDT0000-lwp-OST0002_UUID 5

Trying mount from my 2.1.3 client:

[root@austin7 ~]# mount -t lustre falcao0@tcp:/falcaofs/ /falcaofs/

Client's log:

LustreError: 3675:0:(mgc_request.c:275:config_log_add()) logname falcaofs/-client is too long
LustreError: 15b-f: MGC10.5.0.1@tcp: The configuration from log 'falcaofs/-client'failed from the MGS (-22). Make sure this client and the MGS are running compatible versions of Lustre.
LustreError: 15c-8: MGC10.5.0.1@tcp: The configuration from log 'falcaofs/-client' failed (-22). This may be the result of communication errors between this node and the MGS, a bad configuration, or other errors. See the syslog for more information.
LustreError: 3675:0:(llite_lib.c:961:ll_fill_super()) Unable to process log: -22
Lustre: Unmounted falcaofs/-client
Lustre: 1596:0:(client.c:1796:ptlrpc_expire_one_request()) @@@ Request sent has timed out for slow reply: [sent 1366641507/real 1366641507] req@ffff88003e5a4400 x1432758140272642/t0(0) o250->MGC10.5.0.1@tcp@10.5.0.1@tcp:26/25 lens 368/512 e 0 to 1 dl 1366641512 ref 1 fl Rpc:XN/0/ffffffff rc 0/-1
LustreError: 3675:0:(obd_mount.c:2198:lustre_fill_super()) Unable to mount (-22)

Maybe I missed something. Don't hesitate to ask for more information if needed.

Comments

[Lustre Bull]

Just checked with a 2.3.63 and I got the same error. Actually it was silly, just the slash at the end of the filesystem name.

However I cannot mount the client yet. I'll continue the investigation and close this bug if necessary.

[Andreas Dilger]

Looks like the '/' in "falcaofs/-client" shouldn't be there. That pushes the fsname above 8 characters, which is the limit. It would be worthwhile to mount the MGS and dump the config llog to see if this is written to the log, and see how it got there.

[Diego Moreno (Inactive)]

There was a network problem on my conf test.

Actually I see no problem mounting 2.1.3 clients with 2.3.63 servers. You can close this one, sorry.

[Peter Jones]

ok thanks Diego!