Bug 1734299

Summary: ctime: When healing ctime xattr for legacy files, if multiple clients access and modify the same file, the ctime might be updated incorrectly.
Product: [Community] GlusterFS Reporter: Kotresh HR <khiremat>
Component: ctimeAssignee: Kotresh HR <khiremat>
Status: CLOSED NEXTRELEASE QA Contact:
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: mainlineCC: bugs, pasik
Target Milestone: ---   
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of:
: 1734305 1737745 1739436 (view as bug list) Environment:
Last Closed: 2019-08-01 02:59:49 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Bug Depends On:    
Bug Blocks: 1734305, 1737745, 1739436    

Description Kotresh HR 2019-07-30 08:10:58 UTC
Description of problem:
Ctime heals the ctime xattr ("trusted.glusterfs.mdata") in lookup
if it's not present. In a multi client scenario, there is a race
which results in updating the ctime xattr to older value.

e.g. Let c1 and c2 be two clients and file1 be the file which
doesn't have the ctime xattr. Let the ctime of file1 be t1.
(from backend, ctime heals time attributes from backend when not present).

Now following operations are done on mount
c1 -> ls -l /mnt1/file1  |   c2 -> ls -l /mnt2/file1;echo "append" >> /mnt2/file1;

The race is that the both c1 and c2 didn't fetch the ctime xattr in lookup,
so both of them tries to heal ctime to time 't1'. If c2 wins the race and 
appends the file before c1 heals it, it sets the time to 't1' and updates 
it to 't2' (because of append). Now c1 proceeds to heal and sets it to 't1' 
which is incorrect.


Version-Release number of selected component (if applicable):
mainline

How reproducible:
Always

Steps to Reproduce:
1. Create single brick gluster volume and start it
2. Mount at /mnt1 and /mnt2
3. Disable ctime
    gluster volume set <volname> ctime off
4. Create a file
    touch /mnt/file1
5. Enable ctime
    gluster volume set <volname> ctime on
6. Put a breakpoint at gf_utime_set_mdata_lookup_cbk on '/mnt1'
7. ls -l /mnt1/file1
      This hits the break point, allow for root gfid and don't continue on stbuf->ia_gfid  equals to file1's gfid
8. ls -l /mnt2/file1
9. The ctime xattr is healed from /mnt2. Capture it.
    getfattr -d -m . -e hex /<brickpath>/file1 | grep mdata
10. echo "append" >> /mnt2/file1 and capture mdata
    getfattr -d -m . -e hex /<brickpath>/file1 | grep mdata
11. Continue the break point at step 7 and capture the mdata


Actual results:
mdata xattr at step 11 is equal to step 9 (Went back in time)

Expected results:
mdata xattr at step 11 should be equal to step 10

Additional info:

Comment 1 Worker Ant 2019-07-30 08:14:18 UTC
REVIEW: https://review.gluster.org/23131 (posix/ctime: Fix race during lookup ctime xattr heal) posted (#1) for review on master by Kotresh HR

Comment 2 Worker Ant 2019-08-01 02:59:49 UTC
REVIEW: https://review.gluster.org/23131 (posix/ctime: Fix race during lookup ctime xattr heal) merged (#2) on master by Amar Tumballi