[linux-cifs-client] Re: Timestamps not preserved onCIFSmount

Eric B. ebenze at hotmail.com
Tue Mar 11 14:29:44 GMT 2008


> Am Sonntag, 9. März 2008 schrieb Eric B.:
> > > Am Samstag, 8. März 2008 schrieb Eric B.:
> > > > >Am Samstag, 8. März 2008 schrieb Günter Kukkukk:
> > > >
> > > > >just some additional thoughts on your problem i forgot
> > > > >to mention.
> > > > >When the samba server on your storage device is not running
> > > > >the cifs "unix extensions", the windows like smb functions
> > > > >are used - so you can probably been hit by the currently
> > > > >failing implementation, too.
> > > > >
> > > > >Btw - many NAS devices (even still sold today) are running
> > > > >samba-2.x.x. I've been told in many places, that users had problems
> > > > >when using cifs vfs - but smbfs was working (for their limited 
> > > > >needs).
> > > >
> > > > Thanks for the info.  I'm actually trying cifs b/c of problems I am
> > > > encountering with smbfs; the timestamps work properly, but I keep
> > > > getting
> > > > error/warning msgs regarding inability to set permissions and/or
> > > > ownership
> > > > of the files.  For example:
> > > > # cp -a /home/eric/deploy /mount/smbfs
> > > > cp: failed to preserve ownership for `./deploy/checksum.sh~': 
> > > > Permission
> > > > denied
> > > > cp: failed to preserve ownership for `./deploy/checksum.sh': 
> > > > Operation
> > > > not
> > > > permitted
> > > > cp: failed to preserve ownership for `./deploy/deploy-dev': 
> > > > Operation
> > > > not
> > > > permitted
> > > >
> > > >
> > > > I also get the same error msgs if I try rsync instead:
> > > > # rsync -a /home/eric/deploy
> > > > rsync: chown "/mount/bashful/cifs/deploy" failed: Operation not
> > > > permitted
> > > > (1)
> > > > rsync: chown "/mount/bashful/cifs/deploy/.checksum.sh.TDQedW" 
> > > > failed:
> > > > Operation not permitted (1)
> > > > rsync: chown "/mount/bashful/cifs/deploy/.checksum.sh~.KZ1wck" 
> > > > failed:
> > > > Permission denied (13)
> > > > rsync: chown "/mount/bashful/cifs/deploy/.deploy-dev.KweNfI" failed:
> > > > Operation not permitted (1)
> > > >
> > > >
> > > > I figured that those probs might be avoided with cifs, and indeed 
> > > > they
> > > > are,
> > > > but am running into the timestamp issue.  If I could resolve the
> > > > ownership /
> > > > chown problem in smbfs, I'd be happy to stick with that as well.
> > > >
> > > > Any ideas where I might be able to find help about that?  Maybe a
> > > > newsgroup
> > > > or mailing list that would be good for smb issues like this?  I was
> > > > hoping
> > > > to find a resolution for the cifs problem here, but that proves to 
> > > > be
> > > > too
> > > > difficult, I'll revert to looking for a smbfs soln instead.
> > > >
> > > > Thanks!
> > > >
> > > > Eric
> > >
> > > Was thinking about "getting more info" from your
> > > NAS device.
> > > Cifs is writing it's debug output to the syslog, often located
> > > at /var/log/messages.
> > >
> > >
> > > Please do the following:
> > >
> > > 1.) umount that device
> > > 2.) do 'modprobe cifs' (just to make sure that module is loaded)
> > > 3.) do 'echo 3 > /proc/fs/cifs/cifsFYI'
> > > 4.) in a separate console use 'tail -f /var/log/messages' (or similar 
> > > for
> > > your distro)
> > > 5.) use 'mount -t cifs ....' to connect to the device
> > > 6.) use 'cp -p <src> <dest> to just copy 1 file
> > > 7.) lots of stuff should be send to syslog
> > > 8.) email me the new cifs related syslog entries (samba at kukkukk dot
> > > com)
> >
> > Hmmm  - I tried that but couldn't get any logging to work:
> >
> > # /sbin/modprobe -l cifs
> > /lib/modules/2.6.9-34.ELsmp/kernel/fs/cifs/cifs.ko
> >
> > # tail -f /var/log/messages
> > Mar  8 23:59:00 apollo kernel: SELinux: initialized (dev cifs, type 
> > cifs),
> > uses genfs_contexts
> >
> > That's all I get.
> >
> > If I try to do echo 1 > /prof/fs/cifs/cifsFYI, I get the following:
> > Mar  9 00:11:39 apollo kernel: SELinux: initialized (dev cifs, type 
> > cifs),
> > uses genfs_contexts
> > Mar  9 00:11:48 apollo kernel: Status code returned 0xc0000022
> > NT_STATUS_ACCESS_DENIED
> > Mar  9 00:11:48 apollo kernel: Status code returned 0xc0000148
> > NT_STATUS_INVALID_LEVEL
> >
> >
> > however, if I check dmesg, I do get some output with cifsFYI set to 1. 
> > I
> > have emailed you the output.
> >
> > > Btw - are you familiar with network sniffers like wireshark (former
> > > ethereal) ?
> > > Network sniffs could also help to track that down.
> >
> > Hmm - ok - not sure how I can use wireshark to track that down; I am 
> > working
> > remotely (no physical access to the machines), and all machines run 
> > through
> > switches.  And the RHEL box is a production server, which makes it hard 
> > to
> > run any type of network sniffer on it (not really supposed to be 
> > installing
> > new s/w on it, etc).
> >
> > > Btw - using smbfs will only be a temporary solution, cause it will be
> > > dropped
> > > from the linux kernel very soon.
> >
> > I realize that, which is why I am trying to get cifs to work.  But if I
> > can't, then I might need to revert to smbfs...
> >
> > Thanks for the help!
> >
> > Eric
>
> >
>
>
> Eric,
>
> thanks for sending private emails containing your cifs debug logs!  :-)
>
> From the following (and later entries) i can see that the "unix 
> extensions"
> are enabled on your NAS device:
> ...
> Â fs/cifs/connect.c: rfc1002 length(big endian)0x5d)
> Â fs/cifs/connect.c: Security Mode: 0x3 Capabilities: 0x80e3fd Time Zone: 
> 0
> ...
> #define CAP_UNIX               0x00800000
>
> Your cifs version is surely a "bit outdated":

Agreed; it is definitely outdated, as my kernel is several releases behind. 
Of course, the problem I was/am encountering with this particular production 
server is that I cannot just update my kernel to run tests without proper 
planning, notif, etc, or it will cause a lot of down-time disruption.  As I 
posted an another msg, I have since taken a spare maching, installed 
CentOS4.6 on it, with the latest kernel / updates, and the timestamp seems 
to be working fine now.

> Steve, additional info for you:
> echo 1 > /proc/fs/cifs/cifsFYI    is working
> echo 3 > /proc/fs/cifs/cifsFYI    is _not_ sending _any_ debug info at all 
> !?

However, even with this latest version of cifs, echo 3 > 
/proc/fs/cifs/cifsFYI doesn't seem to be working.  I am using the following 
cifs module:
# /sbin/modinfo cifs
filename:       /lib/modules/2.6.9-67.0.4.ELsmp/kernel/fs/cifs/cifs.ko
parm:           CIFSMaxBufSize:Network buffer size (not including header). 
Default: 16384 Range: 8192 to 130048
parm:           cifs_min_rcv:Network buffers in pool. Default: 4 Range: 1 to 
64
parm:           cifs_min_small:Small network buffers in pool. Default: 30 
Range: 2 to 256
parm:           cifs_max_pending:Simultaneous requests to server. Default: 
50 Range: 2 to 256
author:         Steve French <sfrench at us.ibm.com>
license:        GPL
description:    VFS to access servers complying with the SNIA CIFS 
Specification e.g. Samba and Windows
version:        1.48a.RH BA7BA9F45C60D2722E9A895
vermagic:       2.6.9-67.0.4.ELsmp SMP 686 REGPARM 4KSTACKS gcc-3.4
depends:

> Eric -anyway- you should file a bug report to https://bugzilla.samba.org/
> (cifs-client) and add your cifs dmesg there - so others can comment, too.

Is it still worth filing a bug report even though it seems to be working 
now?

Thanks,

Eric





More information about the linux-cifs-client mailing list