Bug 1314622

Summary: [RHEL-7] gluster-swift: During GETs, duplicate fd leaks for files that need Etag recomputed
Product: [Red Hat Storage] Red Hat Gluster Storage Reporter: Prashanth Pai <ppai>
Component: gluster-swiftAssignee: Prashanth Pai <ppai>
Status: CLOSED ERRATA QA Contact: surabhi <sbhaloth>
Severity: high Docs Contact:
Priority: high    
Version: rhgs-3.1CC: nlevinki, rcyriac, rhinduja, rhs-bugs, thiago
Target Milestone: ---Keywords: ZStream
Target Release: RHGS 3.1.3   
Hardware: All   
OS: All   
Whiteboard:
Fixed In Version: swiftonfile-2.3.0-0 Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
: 1328443 (view as bug list) Environment:
Last Closed: 2016-06-23 05:31:38 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: 1299184, 1328443    

Description Prashanth Pai 2016-03-04 05:47:54 UTC 
Description of problem:

A fd was not bring closed after it was duplicated. This code path can
be easily hit when doing a GET on a file that needs Etag (md5sum) to
be recalculated.

# echo "aaaaaaaaaaa" > /mnt/gluster-object/test/c1/f6

# curl -i http://localhost:8080/v1/AUTH_test/c1/f6 -X GET

# strace -ff -p 10245 -e open,close,dup

    Process 10245 attached with 21 threads
    [pid 10245] open("/mnt/gluster-object/test/c1/f6", O_RDONLY|O_CLOEXEC) = 7
    [pid 10245] dup(7)                      = 14
    [pid 10245] close(7)                    = 0

# ls -l /proc/10245/fd | grep f6

    lr-x------. 1 root root 64 Mar  3 16:38 14 -> /mnt/gluster-object/test/c1/f6
Comment 1 Prashanth Pai 2016-03-04 05:50:58 UTC 
Upstream change on review: http://review.gluster.org/13593
Comment 4 Prashanth Pai 2016-03-10 09:11:45 UTC 
Upstream change merged: http://review.gluster.org/#/c/13593/
Comment 9 surabhi 2016-04-25 09:20:41 UTC 
Followed following steps :

1. PUT an object with some data :
curl -i -d "swiftonfile" http://localhost:8080/v1/AUTH_test/c1/file1 -X PUT

2. curl -i http://localhost:8080/v1/AUTH_test/c1/file1 -X GET

3. Observe the process by attaching to strace:

Process 19074 attached with 21 threads
[pid 19074] open("/mnt/gluster-object/test/c1/file1", O_RDONLY|O_CLOEXEC) = 10
[pid 19074] close(10)                   = 0
[pid 19074] close(6)                    = 0


4. # echo "aaaaaaaaaaa" > /mnt/gluster-object/test/c1/file1

[pid 19074] open("/mnt/gluster-object/test/c1/file1", O_RDONLY|O_CLOEXEC) = 10
[pid 19074] dup(10)                     = 12
[pid 19074] close(12)                   = 0
[pid 19074] close(10)                   = 0

5. # curl -i http://localhost:8080/v1/AUTH_test/c1/file1 -X GET

Process 19074 attached with 21 threads
[pid 19074] open("/mnt/gluster-object/test/c1/file1", O_RDONLY|O_CLOEXEC) = 10
[pid 19074] close(10)                   = 0
[pid 19074] close(6)                    = 0
[pid 19074] open("/mnt/gluster-object/test/c1/file1", O_RDONLY|O_CLOEXEC) = 10
[pid 19074] dup(10)                     = 12
[pid 19074] close(12)                   = 0
[pid 19074] close(10)                   = 0
[pid 19074] close(6)                    = 0

ls -l /proc/19074/fd | grep file1

There is no fd leak observed on open , dup and also verified there is close on each open. Marking the BZ to verified.

Versions:
swiftonfile-2.3.0-2.el7rhgs.noarch
glusterfs-3.7.9-1.el7rhgs.x86_64
Comment 11 errata-xmlrpc 2016-06-23 05:31:38 UTC 
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHEA-2016:1289