Fatal Trap 12 when copying files with ACL

Status
Not open for further replies.

dangil

Cadet
Joined
Apr 25, 2014
Messages
3
I am running freenas 9.2.1.5 x64 on a intel serverboard S1200BTL with 16GB of ECC RAM, a xeon E3-1230 v2 and a integrated LSI 1064e controller exposing 4 SAS 1TB Disks (JBOD)

I created a RaidZ1 Pool with all HDs named Pool0

I created two datasets within that Pool0, named E and F

I shared those datasets via CIFS, with Active Directory integration. AD is working fine (wbinfo -t output : "checking the trust secret for domain DOMAIN.local via RPC calls succeeded", wbinfo -u and -g show domain users and groups)

When I start copying 200GB of files to those shares via robocopy from a windows server, with /copy CATS options enabled (to copy all ACL info) after a few hours the whole system crashes with fatal trap 12

after a reboot, at zpool import, the whole system crashes again and the Pool is hosed forever. not even readonly works

Already ran memtestx86 for 24hours without incident.

I am not 100% sure, but I think that if I copy files without copying the ACL, everything works fine

attached the crash textdump and info files

what are the next steps to solving this bug?

EDIT: I've being running Bart's Stuff Test 24-hours straight , writing nonstop data over CIFS share from a windows machine without issues. also, I copied over 400GB of data from windows server via CIFS share with robocopy, but without copying ACLS (just /copy DAT) without issues
 

Attachments

  • crash.zip
    17.9 KB · Views: 213
D

dlavigne

Guest
If you don't get an answer here, it might be worth creating a bug report at bugs.freenas.org and posting the issue number here. The dev might instruct you on how to get a crash report so that the source of the kernel error 12 can be determined.
 

dangil

Cadet
Joined
Apr 25, 2014
Messages
3
I was going to do that, but first I was wandering if the crash report I posted here is what the dev might need
 
D

dlavigne

Guest
It's a good start for the ticket. The dev will let you know if more info is needed.
 

DaVo

Cadet
Joined
Jun 4, 2013
Messages
9
Hi,

I have another crash dump here for an issue which seems to have the same cause.

What I did:
> Create new ZFS volume
> Create dataset in within this
> Create folder in this dataset
> Assign permissions to this folder (real group names replaced):
Code:
setfacl -a 0 g:admin_grp:rwxpDdaARWcCos:fd:allow .
setfacl -a 0 g:public_grp:Rc:fd:allow .
setfacl -a 0 g:downloader_grp:Rc:fd:allow .
setfacl -a 0 g:group1:Rc:fd:allow .
setfacl -a 0 g:group2:Rc:fd:allow .
setfacl -a 0 g:group3:Rc:fd:allow .
setfacl -a 0 g:group4:rxaRcs:fd:allow .
setfacl -a 0 g:group5:Rc:fd:allow .
setfacl -a 0 g:group6:Rc:fd:allow .
setfacl -a 0 g:group7:Rc:fd:allow .
setfacl -a 0 g:group8:Rc:fd:allow .
setfacl -a 0 g:group9:Rc:fd:allow .
setfacl -a 0 g:group10:Rc:fd:allow .
setfacl -a 0 g:gruppe1_grp:Rc:fd:allow .
setfacl -a 0 g:gruppe2_grp:Rc:fd:allow .
setfacl -a 0 g:gruppe3_grp:Rc:fd:allow .
setfacl -a 0 g:gruppe4_grp:Rc:fd:allow .
setfacl -a 0 g:gruppe5_grp:Rc:fd:allow .
setfacl -a 0 owner@:rwxpDdaARWcCos:fd:allow .


> Tried to remove the permissions again:
Code:
setfacl -x 0 .
setfacl -x 0 .
setfacl -x 0 .
setfacl -x 0 .
setfacl -x 0 .


Somewhen between the 5th and 10th command, the system crashed. Dump 0 was the result (I definitively did not remove all entries).

I could reproduce this behaviour a second time - see dump 1 in zip file.

I'm not sure whether this issue belongs to the one described here: https://bugs.freenas.org/issues/5225#change-21712 but maybe it does and then this dump could help...

Best regards

Oliver
 

Attachments

  • crashdumps.zip
    44.3 KB · Views: 201

DaVo

Cadet
Joined
Jun 4, 2013
Messages
9
So it took a little long to write my post.

I'm looking forward to 9.2.1.6 and am curious whether this solves my problem as well...
 

Roger Wilco

Explorer
Joined
Jul 17, 2014
Messages
65
Hi,

<snip>

> Tried to remove the permissions again:
Code:
setfacl -x 0 .
setfacl -x 0 .
setfacl -x 0 .
setfacl -x 0 .
setfacl -x 0 .


Somewhen between the 5th and 10th command, the system crashed. Dump 0 was the result (I definitively did not remove all entries).

<snip>

Oliver

Hi,
can confirm the same behaviour on 9.2.1.7 (dumps are available, if needed), though I tried to remove an ACE at an index which did not exist:

Code:
[root@freenas01] /mnt/datapool/cifs/shares/scanner# getfacl .
# file: .
# owner: nobody
# group: smbusers
        user:canon:-w-p----------:-d----:allow
        user:canon:-w-p----------:f-----:allow
            owner@:-----d--------:f-----:allow
            owner@:----D---------:-d----:allow
            owner@:--x-----------:-d----:allow
            owner@:r-----aAR-cC--:fd----:allow
            group@:-----d--------:f-----:allow
            group@:----D---------:-d----:allow
            group@:--x-----------:-d----:allow
            group@:r-----a-R-c---:fd----:allow
         everyone@:--------------:fd----:allow
        user:canon:-w-p----------:-d----:allow
        user:canon:-w-p----------:f-----:allow
        user:canon:--x-----------:-d----:allow
        user:canon:r-----a-R-c---:fd----:allow
            owner@:------aA--cC--:fdi---:allow
            owner@:------aA--cC--:------:allow
            group@:-----d--------:f-----:allow
            group@:----D---------:-d----:allow
            group@:--x-----------:-d----:allow
            group@:r-----a-R-c---:fd----:allow
         everyone@:----D---------:-d----:deny
[root@freenas01] /mnt/datapool/cifs/shares/scanner# setfacl -x 11 .
[root@freenas01] /mnt/datapool/cifs/shares/scanner# setfacl -x 11 .
[root@freenas01] /mnt/datapool/cifs/shares/scanner# setfacl -x 11 .
[root@freenas01] /mnt/datapool/cifs/shares/scanner# setfacl -x 11 .
[root@freenas01] /mnt/datapool/cifs/shares/scanner# setfacl -x 11 .
[root@freenas01] /mnt/datapool/cifs/shares/scanner# setfacl -x 11 .
[root@freenas01] /mnt/datapool/cifs/shares/scanner# setfacl -x 11 .
[root@freenas01] /mnt/datapool/cifs/shares/scanner# setfacl -x 11 .
[root@freenas01] /mnt/datapool/cifs/shares/scanner# setfacl -x 11 .
[root@freenas01] /mnt/datapool/cifs/shares/scanner# setfacl -x 11 .
[root@freenas01] /mnt/datapool/cifs/shares/scanner# setfacl -x 11 .
[root@freenas01] /mnt/datapool/cifs/shares/scanner# setfacl -x 11 .
setfacl: .: acl_delete_entry_np() failed: Invalid argument
[root@freenas01] /mnt/datapool/cifs/shares/scanner# getfacl .Write failed: Broken pipe


Please do not try to see any sense in the permissions :)
I've added the first 11 entries at index 0-10, and tried to remove those that existed before, through removing the entry at index 11 quite fast, until things stopped and the session closed during reboot.

The pool seems ok, though.

- Roger Wilco
 
Status
Not open for further replies.
Top