Bug ID 1033229
Summary timeout when accessing disconnected mount.cifs resources
Classification openSUSE
Product openSUSE Tumbleweed
Version Current
Hardware x86-64
OS SUSE Other
Status NEW
Severity Normal
Priority P5 - None
Component Samba
Assignee samba-maintainers@SuSE.de
Reporter haenig@cosifan.de
QA Contact samba-maintainers@SuSE.de
Found By ---
Blocker ---

Created attachment 720515 [details]
screenshot of wireshark showing start of disconnect - Tumbleweed

I am using several different Linux boxes (openSuSE 13.2, openSuSE Leap 42.2 and
openSuSE Tumbleqweed) against a Windows 2008R2 Server

mounting and using shares on all systems is done through the same set of
commands, e.g.
# mount.cifs //windows_host/public2 /mnt/public2 --verbose -o
iocharset=iso8859-1,noexec,file_mode=0777,uid=12345,gid=12345,domain=mydomain,credentials=/etc/samba/my_credentials

after not accessing any of the files of the public2 resource (for approx. 15
minutes) the Windows's Server Timeout comes to play and the resource
disconnects (# cat /proc/fs/cifs/DebugData)

whereas the openSuSE Leap 42.2 (as well as the older 13.2) system immediately
reconnects if need arises and one does not notice any "after effects",
applications in the openSuSE Tumbleweed most of the time simply freeze and
after some local timeaout one gets a "resource not available" or similar. 

a comparision of tcpdumps showed some differences:

openSuSE Leap 42.2:
- before the disconnect every 60 seconds there is an SMB Echo Request => SMB
Echo Response handshake
- after the disconnect (until reconnect) every 60 seconds there is a new
connection opened by the client and reset by the server (without any SMB
related data or packets)
- on access to the resource by an application at client side the reconnect
seems instantaneous


openSuSE Tumbleweed 
- before the disconnect every 60 seconds there is an SMB Echo Request => SMB
Echo Response handshake just like with the other system(s)
- but with the disconnect the client initiates roughly 1446 new connections per
second containing an SMB Echo Request for which one sees no response from the
server.
- on access to the resource by an application most of the time one runs againt
another timeout / is not able to access the data resting there - in the
screenshot from tcpdump/wireshark this lasted approximately 12 seconds during
wich time the respective application is of course "frozen".


Versions

# ----------------------------------------
# openSuSE 13.2 - works
# ----------------------------------------
thomas:~ # mount.cifs --version 
mount.cifs version: 6.4

thomas:~ # rpm -qf /sbin/mount.cifs
cifs-utils-6.4-3.4.1.x86_64

# ----------------------------------------
# openSuSE Leap 42.2 - works
# ----------------------------------------

vm_test:~ # uname -a
Linux vm_test 4.4.57-18.3-default #1 SMP Thu Mar 30 06:39:47 UTC 2017 (39c8557)
x86_64 x86_64 x86_64 GNU/Linux

vm_test:~ # mount.cifs --version
mount.cifs version: 6.5

vm_test:~ # rpm -qf /sbin/mount.cifs
cifs-utils-6.5-1.5.x86_64

# ----------------------------------------
# openSuSE Leap Tumbleweed - does NOT work
# ----------------------------------------

thomas:~ # uname -a
Linux thomas 4.10.8-1-default #1 SMP PREEMPT Fri Mar 31 17:16:00 UTC 2017
(ea9dcd4) x86_64 x86_64 x86_64 GNU/Linux

thomas:~ # mount.cifs --version
mount.cifs version: 6.5

thomas:~ # rpm -qf /sbin/mount.cifs
cifs-utils-6.5-3.1.x86_64


You are receiving this mail because: