Losing access to shares after some minutes on Win7 machines

[Elmojo]

So I'm setting up a new unraid server for my company's office.  Everything is working okay, except that their old 'server' as well as several of their desktop workstations are Win7 boxes.  I can connect to the unraid shares just fine...for a while;  maybe 20 minutes or so.  However, when I'm trying to transfer data from the old server to the new one, after a few minutes, I lose connectivity to the new unraid box.  I can still see it in Windows explorer on a Win7 box, but I can't browse it.  I get an error that "the connection has not been restored", or similar.  Oddly, after a while, the connection will come back for a bit.  Any idea what might be going on here?

[itimpi]

You are likely to get better informed feedback if you attach your system’s diagnostics zip file to your next post in this thread. It is always a good idea when asking questions to supply your diagnostics so we can see details of your system, how you have things configured, and the current syslog.

[Elmojo]

I would have, but I didn't know how, until you just linked that helpful guide.

I tried dumping the diags, but it took forever, then gave an 'out of memory' error and failed.

It seemed to be trying to index the entire contents of the cache drive for some reason.  Is that normal?

I'm not really comfortable with it sending a full file list of everything on that drive anyway, since my company does work for the DoD, and there are rules about what data I can and can't release.  I have read the regs closely enough to see if it extends to file names, but I wouldn't be a bit surprised.

[itimpi]

The diagnostics should not be trying to index any drive.   Not quite sure what you are seeing.

 

Do you have Mover logging enabled?   That is the only thing that should result in lots of file names showing up in the logs, and you only want that enabled if investigating why mover is not behaving as you expect as it both stops the logs being anonymised well and generates lots of output.

[Elmojo]

I'm not sure what it's doing either.  I've never seen anything like it.

Yes, mover logging is enabled.  That is the default.  Should I turn it off?

I've tried invoking the mover before running the diags, and don't see any difference.

[ChatNoir]

1 hour ago, Elmojo said:

That is the default.

Unless there was a glitch at some time, I don't think that's the case.

 

As stated by itimpi, it is only to be used when debugging a specific mover issue, not constantly as it will flood your log and prevent proper anonymizing.

[Elmojo]

12 hours ago, ChatNoir said:

Unless there was a glitch at some time, I don't think that's the case.

 

As stated by itimpi, it is only to be used when debugging a specific mover issue, not constantly as it will flood your log and prevent proper anonymizing.

I don't know what to tell you.  I know I never enabled it, since I didn't even know it was there until it was pointed out to me in this thread.

This is basically a brand new install.  I've only done basic setup, installed a couple containers, and started copying data over from my old server.

I'm guessing I should turn it off and try the diagnostic export again?

[itimpi]

5 hours ago, Elmojo said:

I'm guessing I should turn it off and try the diagnostic export again?

Yes, but reboot the server as well to clear the current logs.

[Elmojo]

Mover logging disabled.  Server rebooted.  Ran diags again, much better this time.

aeserver-diagnostics-20240710-2118.zip

[Elmojo]

No one has any ideas? :/

[JorgeB]

Nothing relevant logged that I can see, do you know a time it was showing the problem?

[Elmojo]

Not really.  It's a bit random.  Sometimes the shares will remain available for hours, other times, I'll lose access within a few minutes.  After a while, they will come back.  It's very odd.  I'm pretty sure it's not a general network environment issue, since our old server, sitting right beside the unraid box, and connected to the same switch, has no issues.  The only difference is that's it's running Win7.  Unraid is the only major variable.

[JorgeB]

Honestly that doesn't look like an Unraid issue to me, but see if you can get diags from when the issue is occurring.

[Frank1940]

It has been so long since I used WIN7 that I can't remember what versions of SMB it supports.  I suspect that it is using SMBv1 and NetBIOS...  This was often an issue with NetBIOS of servers disappearing. 

 

You can read a bit about how things work under SMBv1 in this old MS document:

 

   https://learn.microsoft.com/en-us/previous-versions/windows/it-pro/windows-server-2003/cc737661(v=ws.10)?redirectedfrom=MSDN

 

Back in those days, it was found that leaving all of the servers online 24-7 and setting up SMB on one of them with a global parameter so that it was always elected Local Master resulted in the most stable performance.  Adding this to the SMB Extra setting on Unraid will help do this. 
 

[global]
preferred master = yes
os level = 255

 

[Elmojo]

1 hour ago, JorgeB said:

Honestly that doesn't look like an Unraid issue to me, but see if you can get diags from when the issue is occurring.

All due respect, but I don't see how it can be anything but.  The rest of the office has no issues connecting to the old server, or each other.  It's only the 'new' (unraid) machine that has issues.  My definition, that almost has to mean that it's the problem, no?

5 minutes ago, Frank1940 said:

It has been so long since I used WIN7....

 

The Win7 box has no problems connecting to the office.  It's the new unraid machine that is having trouble maintaining a connection to the rest of the office, including the old server.

But to answer your question, the old Win7 server is running SMBv2 and a lightweight sFTP server for our our large format scanner.

[JorgeB]

9 minutes ago, Elmojo said:

that almost has to mean that it's the problem, no?

Not necessarily, but post the diags after the issue and we can see if there are issues logged or not, it's easy to confirm.

[Elmojo]

6 minutes ago, JorgeB said:

Not necessarily, but post the diags after the issue and we can see if there are issues logged or not, it's easy to confirm.

OK, will do, thanks!

[Frank1940]

56 minutes ago, Elmojo said:

But to answer your question, the old Win7 server is running SMBv2 and a lightweight sFTP server for our our large format scanner.

 

You could try this:  SETTINGS    >>>   SMB     and in the 'SMB Settings' section, make sure the two marked setting are as shown:

 

This basically turns off SMBv1 support on the Unraid server. It will force the server to SMBv2 or higher. 

[Elmojo]

3 hours ago, Frank1940 said:

You could try this:  SETTINGS    >>>   SMB     and in the 'SMB Settings' section, make sure the two marked setting are as shown:

Yep, mine looks like that, with the exception that I also have SMB multichannel and the macOS thing turned on.  Otherwise, it's the same.

[Frank1940]

17 minutes ago, Elmojo said:

SMB multichannel

As I recall, this being enabled has given problems in the past...     (I assume that you have macOS computers on the network otherwise its use can cause be problematic.)

[Elmojo]

4 hours ago, Frank1940 said:

As I recall, this being enabled has given problems in the past...     (I assume that you have macOS computers on the network otherwise its use can cause be problematic.)

Oh? Interesting.  I can turn it off, it just seemed like a good idea to have on, for performance.  As for macOS, no, we don't have any Macs in the office.  We do have iPads and iPhones, so I figured it couldn't hurt.  That may have been in error, though.

 

EDIT:  I was wrong, SMB multichannel is OFF.  I had intended to turn it on, but never did.  Only the macOS switch is on.

Edited July 24 by Elmojo
clarification