I have been searching for info on this and haven't found an authoritative answer. From what I have read, oplocks are good because they increase connection speeds, but they are bad because they don't really work, but they actually do work, but they only work in some cases, etc etc. so, here's my problem and my question together: I get tons of these messages every day (over a thousand a day) [2003/06/20 08:19:42, 0] smbd/oplock.c:request_oplock_break(1011) request_oplock_break: no response received to oplock break request to pid 22335 on port 35010 for dev = 2b00, inode = 688540, file_id = 256210 [2003/06/20 08:19:42, 0] smbd/open.c:open_mode_check(652) open_mode_check: exlusive oplock left by process 22335 after break ! For file UHG/Local Settings/Temporary Internet Files/Content.IE5/desktop.ini, dev = 2b00, inode = 688540. Deleting it to continue... is this an indication that I should disable oplocks, or is disabling oplocks a foolish, unsafe thing to do, or is there just some other problem I need to fix to allow me to keep using oplocks? Very confused. -Mark
OK, I don't have a strong understanding of oplocks, but I'm sure someone will correct me where I go wrong. Overgeneralization #1: Disabling oplocks is ALWAYS a safe thing to do. Overgeneralization #2: Oplocks provide a performance boost by allowing the workstation (ws1) to cache a copy of the file locally and set an oplock. This way, the ws1 can assume it has exclusive access and doesn't need to read/write to/from the server for every operation. Occasionally, the ws1 syncs the cached copy with the server copy. When another workstation (ws2) requests access to the file, the server asks the ws1 to break the oplock. Ws1 then syncs the cache with the server, and tells the server that it's released the oplock. The server then tells ws2 it can access the file. If ws1 has the file open for read (not write), ws2 can open the file for read without breaking any oplocks. Overgeneraliztion #3: With oplocks disabled, the workstation must always ask for an exclusive lock before writing to the file, and does not cache a copy. Another workstation can't access the file until the first workstation releases it. Exactly what goes on when things go wrong (server doesn't ask for oplock break; workstation doesn't release oplock, etc.) I can't tell you. As for the meaning of your errors, I haven't a clue. --Jon P.S. -- My philosophy is that if you ask a question and no one answers, tell a lie as gospel truth and everyone will. On 20 Jun 2003, Mark Roach wrote:> I have been searching for info on this and haven't found an > authoritative answer. From what I have read, oplocks are good because > they increase connection speeds, but they are bad because they don't > really work, but they actually do work, but they only work in some > cases, etc etc. > > so, here's my problem and my question together: I get tons of these > messages every day (over a thousand a day) > > [2003/06/20 08:19:42, 0] smbd/oplock.c:request_oplock_break(1011) > request_oplock_break: no response received to oplock break request to > pid 22335 on port 35010 for dev = 2b00, inode = 688540, file_id = 256210 > [2003/06/20 08:19:42, 0] smbd/open.c:open_mode_check(652) open_mode_check: > exlusive oplock left by process 22335 after break ! For file UHG/Local > Settings/Temporary Internet Files/Content.IE5/desktop.ini, dev = 2b00, > inode = 688540. Deleting it to continue... > > > is this an indication that I should disable oplocks, or is disabling > oplocks a foolish, unsafe thing to do, or is there just some other > problem I need to fix to allow me to keep using oplocks? > > Very confused. > > -Mark > >
The original url got me nothing, but this: http://samba.org/~jht/NT4migration/Samba-HOWTO-Collection.pdf got me the doc.>>> John H Terpstra <jht@samba.org> 06/20/03 02:38PM >>>http://samba.org/~jht/NT4migration/Samba-HOWTO-Collection.html
My comment/question leans towards Windows here.... I cant turn oplocks off in Windows (to my knowledge). Therefore they are always ON. If this is the case, And they dont cause any problems, then WHY do we have to turn them off in Samba? Do they not work properly? -----Original Message----- From: Mark Roach [mailto:mrroach@okmaybe.com] Sent: Friday, June 20, 2003 2:40 PM To: John H Terpstra Cc: samba@lists.samba.org; Jonathan Johnson Subject: Re: [Samba] ok, so oplocks: good or bad? On Fri, 2003-06-20 at 15:38, John H Terpstra wrote:> On Fri, 20 Jun 2003, Jonathan Johnson wrote: > > > OK, I don't have a strong understanding of oplocks, but I'm sure someone > > will correct me where I go wrong. > > Those interested in the whole OpLock story might benefit from reading > chapter 14 'File and Record Locking" of: > > http://samba.org/~jht/NT4migration/Samba-HOWTO-Collection.html > > From this coverage it should be obvious that file locking affects a > complex interaction of Client and Server protocols and configuration > settings. Please draw your own conclusions.Hi, John. I'm pretty sure I "get" what oplocks are for and why they are good, I guess my question would be more along the lines of "do they work properly in samba?" along with the error message that prompts the question. I think the question could be further distilled to "Is this an example of oplocks not working properly, or is it something else entirely?" I hope my question makes more sense worded that way. Thanks very much, Mark Roach -- To unsubscribe from this list go to the following URL and read the instructions: http://lists.samba.org/mailman/listinfo/samba
On Fri, Jun 20, 2003 at 02:49:48PM -0500, Brandon Lederer wrote:> My comment/question leans towards Windows here.... I cant turn oplocks off > in Windows (to my knowledge). Therefore they are always ON. If this is the > case, And they dont cause any problems, then WHY do we have to turn them off > in Samba? Do they not work properly?Yes you can turn them off in Windows, there are registry entries to do so on both client and server (as in Samba) - although not per share or wildcard match as you can in Samba. Several Windows application vendors (usually multi-user db vendors) recommend turning them off Windows to Windows, no Samba involved. "Oplocks: nice idea, shame about the Windows client implementation". Jeremy.
On Fri, 20 Jun 2003, Troy.A Johnson wrote:> The original url got me nothing, but > this: > > http://samba.org/~jht/NT4migration/Samba-HOWTO-Collection.pdf > > got me the doc.Apologies for the typo. - John T.> > >>> John H Terpstra <jht@samba.org> 06/20/03 02:38PM >>> > http://samba.org/~jht/NT4migration/Samba-HOWTO-Collection.html > >-- John H Terpstra Email: jht@samba.org
Hi, I disabled oplocks after weeks of corrupt files during network renders amongst 80+ cpus using XP Pro. Disabling oplocks fixed my corrupt file issue however I suspect that my choice of using XP Pro in a serious production env was a bad one. I am now switching to 2K and have a feeling that with oplocks re enabled, file corruption may not occur. Bri- __________________________________ Do you Yahoo!? SBC Yahoo! DSL - Now only $29.95 per month! http://sbc.yahoo.com
On Sun, Jun 22, 2003 at 09:27:24AM -0700, bkrusic@yahoo.com wrote:> Hi, > > I disabled oplocks after weeks of corrupt files during > network renders amongst 80+ cpus using XP Pro. > > Disabling oplocks fixed my corrupt file issue however > I suspect that my choice of using XP Pro in a serious > production env was a bad one. I am now switching to > 2K and have a feeling that with oplocks re enabled, > file corruption may not occur.Indeed - search in the archives for the post I sent with the Microsoft hotfixes for Windows XP SP1 oplock corruption problems. Jeremy.