BIMcloud warnings are getting out of control

[Patrick M]

here is a small sampling of the t/w warnings I have been trouble shooting the past week. This feels like something is broken... or theres something else at play here (my tinfoil hat is on pretty tight right now). And these are about 1/4-1/2 of what I've been sent:

 

  

  

  

 

[Patrick M]

I have a really hard time recommending to designers to 'just keep working and hope it allows a s/r later'. If they end up putting in 2-3 hours and get this message, then put in another 5 hours, thats 8 hours of potential work that needs to be copy/pasted into a file from a .pln; and much of the work can't be just copy/pasted over.

[torben_wadlinger]

You've got me wrong. It is not about saving into a pln. It is just about saving your daily work by clicking onto the Save-Button. AC saves the changes into the local cachefile from which it recovers the data, when AC has crashed or you're working offline. If you can't send your data just save them! After closing AC after work and opening it again the next day you can choose between open the saved file via start window or via teamwork browser. Now comes the best part: if you open the project via teamwork browser and there are changes made from a team mate, AC asks you to send your changes you saved locally the day before! For that reason there is no data or work lost.

Now, as you know, AC licences and TW licences are different. Is there the same number of AC and TW licences available or are there less TW licences?

[torben_wadlinger]

Daily reminder; just complaining doesn't solve the problem. You've got a lot of support from other users. To find the problem it is necessary to go step by step from the client to the server to the router, looking at every interface, every service or daemon, cable or physical connection. Sometimes strange things are happening, like IP-changes because the dhcp is crashing and restarting or is mal-functioning. As we don't know how your network and the internal services are set up there is no chance in giving you any good advice.

From my experience it is not the bimcloud service. I had similar errors when I switched from macOS to Windows 10 Pro (I host every bimcloud as a single vm on esxi). In the end I had to deactivate energy saving in general and in the (virtual) nic driver settings. My best advice therefore is:  check your network, starting with the drivers, through the cables, to the server hardware and the services. Check the log files and look for errors, IP changes, crashes, restarts, time-outs - well, everything.

[jamesMorris]

Sorry @torben_wadlinger, but that is just victim blaming and gets us nowhere. 

Small reminder of where we at:

- For some reason even GS seems unable to understand, an error pop up gets triggered

- According GS support, there are NO errors in the logs

That is part one that is 100% on GS and not on the user. 

Part 2 is that there is really a problem on the users end. 

We hunted down every single issue possible and found nothing. New PC, new AD user, new Bimserver everything!

And even if the issue is on our end, that is not excuse to why ArchiCAD is unable to send the saves during the second or third try. Sure this issue can pop up, when I am on mobile using a VPN and there is a hiccup. But then it should not stay in that broken state. 

[torben_wadlinger]

Nope Sir, I am definitely not blaming the victim. All I say is that there are plenty of other things that can be wrong while the bc is running fine. That's why I asked for more information about the network and the servers and everything. I think I can really help because I have 30+ years experience in IT, Archicad, Bimcloud and all the stuff above, under, left, right, in front of and behind. And diagonally. And mac, windows, linux (and dos). And Architecture. Man, you can divide me in half and you get two high skilled persons. I'm such a great guy!

Seriously, you can send me a pm or we can have a meeting over teams or somethin similar.

[jamesMorris]

Yes you are. Like I said, even if something else is wrong (which I don't deny the possibility),

the MAIN issues are still that ArchiCAD falls into an error state it never can recover, GS does not know what triggered that and there are no logs. 

This has all nothing to do with the users setup. This is 100% on GS. Trying to shift the blame away, is victim blaming. 

Regarding your experience, it is basically the same for me here. 

So I am fine with discussing it here, instead of taking it to a private place like teams. 

What do you wanna know?

[torben_wadlinger]

Let's start with the basics:

1. Server OS
2. LDAP or AD
3. if bc is running as vm, what is the configured virtual hardware
4. what are the energy saving configurations (server, bc), especially of the nic
5. network topology (switch(brand, managed, unmanaged), 1 or 10 GBit, router, dns, dhcp), configuration
6. network cable (cat 6, 6e, 7)

How is the bc configured (IP, domain name)?

Are all bimclouds hostet on the same machine or are they running seperatly (as vm's)?

In case of vm's: does the Server suspend the vm's?

Can you ping the bc-server when the error occurs?

Is the bc web interface working when the error occurs?

Is the bc licence active and available when the error occurs (have a look from a different computer at the licence tab)?

How is the client/server IP resolved (DHCP, manually)? Does it change?

What is the client/server IP when the error occurs?

Have you checked that all IP's are unique?

Have you checked the switch e.g. collission errors (if managed)

Have you checked the nic hardware (I once had random errors because the thermal paste was to old!)

If the bc is running as vm, try to

1. change the physical nic of the virtual nic
2. passthroug a physical nic to the vm

Does the errors occur when the bc Firewall is off? If not, check the rules on the client and the server

If the bc-users are synced by LDAP/AD: is the password policy correct? (I had this problem when upgrading to v28. I was to able to login with my LDAP-password but was thrown out some moments later. After changing my password to a more complex one everthing was ok again)

Have you testet local bc-users (not LDAP/AD)? Do they have similar problems?

I ask all this stuff, because these were things I hunted down when I had trouble with my own bimcloud (and the bimclouds of colleaques). In the early beginning bc had problems with LDAP, but over time I could solve it (with big help from GS from Munich). All sending/receiving problems disapeared when I used a reverse proxy server for https and had the http response time set to 1 hour (3600 sec).

You don't have to report the results of your findings to me, but this should really help you to find the single one issue that causes all the trouble.

[jamesMorris]

Ok, let's do this.

1. Windows Server 2019

2. Bare metal

3. n/a

4. nothing. This is a server Mobo and NIC (Supermicro). 

5. Used to be a managed Cisco Switch, is now a managed MikroTIK Switch. Error happened on both. 

6. Cat7 (and of course switched to rule out a potential issue there). 

No matter if connected by static IP or domain name.

It is only one bare metal machine. 

Issue happened on both, old and new server. 

Yes I can ping the BC server when the error occurs, also access the webGUI.

Yes, other users can see the license server when the error happens. 

Client IP is probably not important here, but yeah it is DHCP. 
Server is static.

Ohh and btw, also applies to IPv6. 

Yes of course the server IP is unique.

Switch shows no error. 

BTW also no errors in the Wireshare logs I sent them.

Not only did we change the NIC but the whole server. 

Windwos Firewall is just default settings. Also does not apply, since this would be a never or always situation, not a sometimes situation. 

No LDAP, just local users on BimCloud Basic

Everything only local and over http.

As you can see, I already played this game with GS. But they still could not find anything. Heck, they could not even tell me what triggered the error. They could not even find something in the logs. 

Now what? Still some kind of user error? As I already said, even GS was running out of stuff to blame on me. Maybe now is the time to take a look in to the mirror. Maybe, just maybe, this isn't the users fault but, and I know that might sound extreme, maybe it is an BimCloud Bug. 

[torben_wadlinger]

Turn of ipv6.

I forgot: it's Windows! In this case: prefer IPv4 over IPv6.

I thougt it over and now I am sure it is a IPv4/IPv6-problem. You didn't answer the dns / dhcp-question. I have a separate machine serving only dhcp, dns and openVPN (pfsense) and IPv6-dhcp ist turned off because of trouble with other services.

Or even better: turn ipv6 off on the switch. In this case you can leave eveything like it is and the "filtering" is done by the switch. And give the switch some time to learn the new routes or reset the routes if possible.

[jamesMorris]

Tried dual stack and IPv4 only already. 

To further answer your questions:

A Windows Server (that is another machine) is doing DHCP/DNS.

OPNsense does no RA, only announces gateway. 
Not that any of this matters, since clients are using the static IPv4 anyway (and not a DNS record) to connect to the BIM server. 

Side note; "just turn off IPv6" is a bad advice old IT "pros" give that are too lazy to learn something new.