Hi Joyce, Thanks for your reply.
On Apr 13, 2010, at 7:39 PM, joyce mcintosh wrote: > These are change notification messages to Explorer to tell it that > something has changed and it needs to refresh. They should occur > when, for example, a new file is created. Explorer requests such > notifications. When you shut down Explorer it stops requesting > these notifications and hence you stop seeing the traffic. > > Joyce > I see. Thanks for clearing this up. But this does not explain why I get 10.000 of these requests every second, at the same second that I enter the share. Nothing on the server is being written/updated that much, and this is happening around the clock, in any directory on the server. The server is not that busy, and this starting happening from day to day. I tried creating a "temp" folder, moving all folders and files to that folder in the same share. It continued to happen even after I moved all the data there, even though no applications or clients should be able to read/write to anything. Is there any way to enable some kind of verbose debugging to see what is actually happening here? kind regards, Kasper > > On 04/13/10 06:19 AM, Kasper Bræmer-Jensen wrote: >> >> Hi, >> >> First post here, but I've been lurking zfs-discuss the past 3 months. >> >> I'm having a very strange problem with one of my CIFS shares. Other shares >> are working just perfectly. When I browse my share from a Windows XP (x64, >> latest SP and updates), i instantly see some very odd traffic that starts >> pretty much instantly. As soon as I close down Explorer or goes to another >> share, the traffic stops and everything works fine again. >> >> I see around 5-10Mbit of traffic, and thousands (~10K/second) of tiny small >> packets are transmitted back and forth, for every workstation/client. This >> is generating a lot of interrupts on the fileserver, raising the load, all >> for no obvious reason. >> >> When I first saw this happening, I thought it could be some kind of indexing >> thing going on, or generating some media preview of the data on the server, >> or similar. But after looking at packets (and considering the huge amount of >> packets getting transferred), I think it might be some kind of CIFS bug. >> >> A quick video of what's happening is here: >> http://kapsel.dk/asdf/strangeTraffic2.mov >> >> and a dump of some of the packets, grabbed on the Windows client >> (192.168.0.61) is here: >> http://dpaste.com/hold/183223/ >> >> The server is an OpenSolaris (snv_133) with "zfs set sharesmb=on pool/fs" >> set. >> >> Any ideas on what I could do about this? This started happening out of >> nowhere, from day to day, and it's somewhat a deal-breaker for us, for >> running OpenSolaris. Currently, with ~8-10 users browsing shares, I see >> around 100K interrupts per second on the server. >> >> kind regards, >> Kasper >> >> _______________________________________________ >> cifs-discuss mailing list >> cifs-discuss@opensolaris.org >> http://mail.opensolaris.org/mailman/listinfo/cifs-discuss >> >
_______________________________________________ cifs-discuss mailing list cifs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/cifs-discuss