[LU-7565] (llite_lib.c:2309:ll_prep_inode()) new_inode -fatal: rc -2 Created: 16/Dec/15  Updated: 29/Jan/19  Resolved: 24/Jan/17

Status: Resolved
Project: Lustre
Component/s: None
Affects Version/s: None
Fix Version/s: None

Type: Bug Priority: Critical
Reporter: Frank Heckes (Inactive) Assignee: WC Triage
Resolution: Cannot Reproduce Votes: 0
Labels: soak
Environment:

lola
build: tip of master (commit ae3a2891f10a19acf855a90337316dda704da5d)


Severity: 3
Rank (Obsolete): 9223372036854775807

 Description   

The error happens during soak testing of build '20151214' (see https://wiki.hpdd.intel.com/display/Releases/Soak+Testing+on+Lola#SoakTestingonLola-20151214)

During normal soak operations (no fault injected) approximately 5% of the test (batch) jobs crash.
The events can be correlated to the error message

Dec 16 02:40:45 lola-31 kernel: LustreError: 110820:0:(llite_lib.c:2309:ll_prep_inode()) new_inode -fatal: rc -2
Dec 16 02:40:45 lola-31 kernel: LustreError: 110820:0:(llite_lib.c:2309:ll_prep_inode()) Skipped 1 previous similar message

on a Lustre client node executing the batch job.
No other message on OSS or MDS nodes can be correleated.



 Comments   
Comment by Cliff White (Inactive) [ 24/Jan/17 ]

Issue did not reproduce- closing

Comment by Alex Zhuravlev [ 28/Jan/19 ]

I'm hitting this locally (very rare).. it seems the root cause is that failed mkdir can't cleanup the name sometimes, then subsequent lookup/stat/whatever finds that in broken state (the name and one stripe exist, but another stripe doesn't).

Comment by Alex Zhuravlev [ 29/Jan/19 ]

can be reproduced with ZFS and ldiskfs

Generated at Sat Feb 10 02:09:59 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.