[LU-9871] Client failures in soak: llite_lib.c:2303:ll_prep_inode()) new_inode -fatal: rc -12 Created: 11/Aug/17 Updated: 28/Mar/18 |
|
| Status: | Open |
| Project: | Lustre |
| Component/s: | None |
| Affects Version/s: | Lustre 2.11.0 |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Major |
| Reporter: | Cliff White (Inactive) | Assignee: | WC Triage |
| Resolution: | Unresolved | Votes: | 0 |
| Labels: | dne2 | ||
| Environment: |
Soak performance cluster version=2.10.51 |
||
| Attachments: |
|
| Severity: | 3 |
| Rank (Obsolete): | 9223372036854775807 |
| Description |
|
Seeing frequent client failures, causing client to be marked unready by soak. [ 5542.452556] LustreError: 23649:0:(namei.c:88:ll_set_inode()) Can not initialize inode [0x280002347:0xf8fc:0x0] without object type: valid = 0x100000001 [ 5542.616751] LustreError: 23649:0:(llite_lib.c:2303:ll_prep_inode()) new_inode -fatal: rc -12 [ 5833.710623] LustreError: 23649:0:(namei.c:88:ll_set_inode()) Can not initialize inode [0x280002342:0x10303:0x0] without object type: valid = 0x100000001 [ 5833.728798] LustreError: 23649:0:(llite_lib.c:2303:ll_prep_inode()) new_inode -fatal: rc -12 I see no other related messages or errors. When I look at an impacted client, memory appears fine. Client jobs wedged, and have to be killed, or die due to slurm timeout. Lustre-log from impacted client attached |
| Comments |
| Comment by Oleg Drokin [ 09/Mar/18 ] |
|
now that I enabled DNE in my test cluster, I see it too. |
| Comment by Andreas Dilger [ 23/Mar/18 ] |
|
I found this during sanityn.sh cleanup for https://testing.hpdd.intel.com/test_sets/42f3c63e-2dc6-11e8-b3c6-52540065bddc in the client console log https://testing.hpdd.intel.com/test_logs/43609e08-2dc6-11e8-b3c6-52540065bddc/show_text |
| Comment by Saurabh Tandan (Inactive) [ 28/Mar/18 ] |
|
Another instance on 2.11RC2 for racer test_1. https://testing.hpdd.intel.com/test_sets/5ee90de4-3268-11e8-b6a0-52540065bddc |