On 3/20/2019 4:42 PM, Ian Branch wrote:
> All apps are located on the physical server, (64bit Win 2012 R2 server) as is the data.
> Apps - C:\AppDirectory.
> Data - C:\AppDirectory\Data

OK

> Most users call the apps via shortcuts to a mapped drive on their wkstn.

In theory i would rate these most likely candidates for weird issue - on
top of edb you might have exe loading issues if network glitches (afaik
exe is not all loaded into ram and windows might page in more code later
from exe).

> Some users log into the physical server via RDP and run the apps on the server.

Should be OK - worst case rdp disconnects but app would continue to run
on server and one can reconnect rdp session.

> There is no chance for there to be apps, different or otherwise, to be running in mixed mode.  There are 18 Apps in all of which 5 are key Day-to-Day apps used constantly.  The others are specialized function apps.

Any chance some of the users have copied app and config locally to their
PC so any change to central config has any effect ?


> They have been running in Local (File share) mode since late 2016 as the Client didn't want to 'Risk' Remote mode.

They are basically running remote since it's file share. Historically
this was usable on Win XP and such (though my experience in DBISAM
mostly) but starting with new windows version this is something you
really want to deprecate if you can.


> Ever since we went live in late 2016 we have been plagued by these very intermittent, loss of connect, couldn't find a file, apparent corrupted record, hiccups.  It would affect several different Apps at the same time however they may not notice if they weren't actually doing anything in the App at the time.I blamed the network, the Client naturally blamed the Apps.

Again my actual experience goes while back but you also might have
things like windows simply timing out the network share connection -
note that windows explorer will auto reconnect if you cloick on maoped
drive or share but same is not always true for apps unless they
specifically handle this.


> I became convinced it was a network issue when an app, running on the physical server via rdp, that simply uses a standard windows API call to see what files, app & data, are in use, failed.  It polls/updates every 5 seconds.
> It failed with a network connection error, at the same time as Users experienced a hiccup in their Apps.

Was that app on the same server and did it use "c:\AppDirectory.."
syntax or file share syntax ?

If latter then yes looks like network connection issue or such - things
like local IP or share name can become briefly invalid if network say is
disconnected for few seconds (if you pull cable and plug back in for
example).

Windows event log might log something in cases where Windows network
stack detects an issue like connection disconnected etc.

> Anyway, it has all been running 100% since yesterday morning and the Users are loving it.

Great!

> That's it.  Sorry for the long tale, hope I didn't bore you, but it was good to document.

Thanks - always interesting get some background