Delayed Write Failed when saving over a network

[indiegirl]

My school runs a Win2k3 server backend, with one domain. We have a 'shared' area for staff to save to, and individual 'home' areas as well.

 

Two members of my staff (both in the same office) are reporting problems saving to the shared area. They tell me that they get a "delayed write failed" error message, with Windows being unable to save the .tmp file - so it can't actually save the .doc (in Word, for example).

 

One of the two is also unable to save to her 'home' area.

 

The problem is that this is intermittent - sometimes it will work, others it won't. The machines are all XP sp2 with all updates, and Office 2k3 with all updates. We thought that the updates had solved the problem this morning, indeed - both staff members could save to the shared area, although 3 hours later they reported that they couldn't. Another staff member in the same office is able to save to both areas problem-free.

 

Now personally I think this could be a network lag, or certainly problems with the network cabling somewhere. My Network Manager (I'm a lowly technician) doesn't agree. Am I barking up the wrong tree? Has anyone else ever had the 'delayed write failed' error before?

[ChrisH]

Ive seen that error before and in most cases it was usually before one of my disks popped its clogs. This was on a local machine. Ive had it twice on my home machine and have seen it on a few work stations here before they go into melt down.

[indiegirl]

Interesting. The main area we're saving back to (on our 'file server') is a brand new server where all staff and students have access. No other member of staff have had that error message when saving to the same shared area.

 

Both local machines are also brand new.

 

I know it's possible, but the likelihood of the server disk being about to go, or both local machine ones seems remote to me.

[disinfo]

Yes I agree with ChrisH,

 

this usually inicates a hardware faliure on the Hard Drive itself, you will see this message if you pull the plug on an external Hard Drive while a file transfer is taking place.

[ChrisH]

If you think it's the network then you need to test it using something like Ethereal to see if theres anyting odd about the traffic getting slowed down or something.

[Dos_Box]

Do you have roaming profiles, a major cause of network congestion?

[indiegirl]

I'm definitely going to test out the network (next week, when it's much quieter).

 

We don't use Roaming Profiles (god, no!).

 

I was just interested to see what others might suggest. I'll try another machine as well in that office to see if the same problem occurs. It's just very strange that two out of three machines in the office would suffer...

[Ric_]

You don't have mice do you (the squeeky type - rather than the computer type)?

 

Mice (or rats) can be a cause of network disruption when the little blighters decide to feast on your Cat5.

[ChrisH]

Try plugging them in a different switch or at least switch ports if you can.

[kingswood]

We had this error for some time. The OS writes data to cache and then when it needs to read it goes hunting for it- but doesnt find it! But it can also be (as others have pointed out) hardware related (even though your drives are new it isn't unusual as you'll know to get a bad one).

 

Also, check that your network connections between client and server aren't going down intermittently (we had students in sixth form turning off a switch between one area of the school and the server...till we found the key for the cupboard ;-)

 

Apparently Windows XP SP2 should have fixed this error if it's client-side, so if you're running SP2 look elsewhere. And back up your server :-)

 

Paul

[eejit]

We've had this problem in the past too, looking on line there are loads of people with it. As far as I remember, people usually reported that finding updated drivers for their NICs solved it, but for us disabling DMA on the harddrives solved it.

 

We only got the error when copying large files >100MB.

[indiegirl]

ChrisH: I'll be plugging them into different switches if I can

 

kingswood: I suspect that as these points go back to the Sixth Form cabinet, we could have the same problem as you . Both the machines are XP SP2, so I don't see any errors on that side. I would've thought that if it was a disk going, everyone who writes to that 'shared area' (which is the entire staffing lot - teaching and support) would have problems - which they don't.

 

eejit: we get it when saving small Word docs! I will check for NIC driver updates though - just to see if that helps (and I don't think we have DMA on but some 'thoughtful' bugger (read my manager) might've put that on)...

 

Thanks all for your suggestions so far, I'll probably report back once troubleshooting has taken place!

[ajbritton]

Look in the event logs on the workstations and servers at around the time of the error. If the problem is client side, or network related then there may be messages to that effect on the workstation. If there are problems with the hard disk on the server then there may be info in the server event log.

 

Other than that, general troubleshoot advice applies. Change one thing at a time. If possible, swap the entire PC out and see if the problem goes with the PC or stays at the network point. You get the general idea!

[indiegirl]

We *think* we've solved the problem here. It was indeed network related...:

 

It appeared that some clever sod had daisy-chained two of the three office network points together (coincidently the two which the PCs had the DWF errors on). Coupled with the fact that these were only running over Cat5e back to the switch, it created a painfully slow lag which caused the errors.

 

The points have been recabled, and fibre's gone in, and wahey... no more problems :-)

 

(just thought I'd update you all)

 

Thanks for your thoughts.

[Dos_Box]

I'm glad it's sorted out. Who on Earth did the cabling!!

[indiegirl]

Good question! I don't actually know - only that it was done a very, very long time ago, probably by a cheap firm!