[linux-cifs-client] Re: problems with cifs

[Gorazd Golob]

> Forgot to ask (it may be obvious from one of your earlier posts but I am
> not certain).
>
> You said that this is much improved but that mount locks from one
> process (can not chdir etc.) but they start working when another proc
> accesses it -
After some time (~10min). Real situation .. Some file are coping from
mount point and coping stops at sudden (no error, just no new data in
file - cp command stall) You want to chdir to mount point and you can't
.. after command there is a long long delay. Even if you try something
with df (I don't know from where df got info) there is a long delay
(usually I do break before it ends or it ends in ~10min as I said with
normal access to mount point directory or lower into it..).
In /proc/fs/cifs/DebugData you can see :

Display Internal CIFS Data Structures for Debugging
---------------------------------------------------
Servers:

1) Name: 192.168.x.y  Domain: NOVFORUM Mounts: 1 ServerOS: Unix
        ServerNOS: Samba 3.0.7  Capabilities: 0x80e3fd
        SMB session status: 3   TCP status: 1
        Local Users To Server: 1 SecMode: 0x3 Req Active: 0
MIDs:


Shares:

1) \\server4\xxxx Uses: 1 Type: NTFS Characteristics: 0x0
Attributes: 0x3
PathComponentMax: 255 Status: 3 type: 0         DISCONNECTED

(there are 4 like this.. al the same - from server4-server1) .. And full
snip from Stats (both snips are also in one of previous email on the list)

Resources in use
CIFS Session: 4
Share (unique mount targets): 4
SMB Request/Response Buffer: 4
SMB Small Req/Resp Buffer: 0
Operations (MIDs): 0

4328 session 28 share reconnects
Total vfs operations: 40082984 maximum at one time: 2

1) \\server4\xxxx DISCONNECTED
SMBs: 9882909 Oplock Breaks: 0
Reads: 9875331 Bytes 0
Writes: 0 Bytes: 0
Opens: 83 Deletes: 0
Mkdirs: 0 Rmdirs: 0
Renames: 0 T2 Renames 0
2) \\server3\xxxx DISCONNECTED
SMBs: 9718022 Oplock Breaks: 0
Reads: 9710364 Bytes 0
Writes: 0 Bytes: 0
Opens: 96 Deletes: 0
Mkdirs: 0 Rmdirs: 0
Renames: 0 T2 Renames 0
3) \\server2\xxxx DISCONNECTED
SMBs: 9742057 Oplock Breaks: 0
Reads: 9733958 Bytes 0
Writes: 0 Bytes: 0
Opens: 88 Deletes: 0
Mkdirs: 0 Rmdirs: 0
Renames: 0 T2 Renames 0
4) \\server1\xxxx DISCONNECTED
SMBs: 10197951 Oplock Breaks: 0
Reads: 10189623 Bytes 0
Writes: 0 Bytes: 0
Opens: 151 Deletes: 0
Mkdirs: 0 Rmdirs: 0
Renames: 0 T2 Renames 0




>
> Could you verify which ops are pending - when the process is not working
> (no chdir etc.) - by looking at /proc/fs/cifs/DebugData checking whether
> that process has a pending request (mid) and let me know what command it
> is and the total number of pending requests are at that moment.  I want
> to know which operation is hung (if any) - and if it is as simple as
> somehow chdir does not cause network traffic because it calls
> lookup/revalidate/getattr and still "remembers" the -ENOENT or -EIO
> until the revalidate cache time expires (I think you can disable inode
> metadata caching via /proc/fs/cifs - and force revalidate on all lookups
> if that is the case - so I am wondering if what is happening is that you
> get an error on a lookup and for the next 30 seconds you will continue
> to get errors on lookup of that file - perhaps it is some sideeffect of
> the kernels dentry caching & negative dentries).
>
>
> On Thu, 2004-12-09 at 08:18, Gorazd Golob wrote:
> > Dear Steve!
> >
> > There is a big progress in cifs support for linux kernel  over last time
> > and I still have problems using it.
> >
>
>