Network keeps resetting

[ajgringo619]

I found references to this bug starting with VMware Player v8, so I'm not exactly sure if this is the same problem. I've been noticing a resetting of my network every 10-15 minutes. Here is a sample of what happens (this is 10 minutes after a reboot):

Spoiler

Aug 06 19:01:24 NetworkManager[1965]: <info> [1628301684.0658] manager: NetworkManager state is now CONNECTED_SITE
Aug 06 19:01:24 vmnetBridge[2412]: RTM_NEWROUTE: index:2
Aug 06 19:01:24 vmnetBridge[2412]: RTM_DELROUTE: index:2
Aug 06 19:01:24 vmnet-natd[3207]: RTM_NEWADDR: index:2, addr:192.168.0.15
Aug 06 19:01:24 vmnet-natd[3207]: RTM_NEWROUTE: index:2
Aug 06 19:01:24 vmnet-natd[3207]: RTM_DELROUTE: index:2
Aug 06 19:01:24 dbus-daemon[1964]: [system] Activating via systemd: service name='org.freedesktop.nm_dispatcher' unit='dbus-org.freedesktop.nm-dispatcher.service' requested by ':1.12' (uid=0…bel="unconfined")
Aug 06 19:01:24 systemd[1]: Starting Network Manager Script Dispatcher Service...
-- Subject: A start job for unit NetworkManager-dispatcher.service has begun execution
-- Defined-By: systemd
-- Support: http://www.ubuntu.com/support
--
-- A start job for unit NetworkManager-dispatcher.service has begun execution.
--
-- The job identifier is 3825.
Aug 06 19:01:24 dbus-daemon[1964]: [system] Successfully activated service 'org.freedesktop.nm_dispatcher'
Aug 06 19:01:24 systemd[1]: Started Network Manager Script Dispatcher Service.
-- Subject: A start job for unit NetworkManager-dispatcher.service has finished successfully
-- Defined-By: systemd
-- Support: http://www.ubuntu.com/support
--
-- A start job for unit NetworkManager-dispatcher.service has finished successfully.
--
-- The job identifier is 3825.
Aug 06 19:01:24 NetworkManager[1965]: <info> [1628301684.2437] manager: NetworkManager state is now CONNECTED_GLOBAL
Aug 06 19:01:24 vmnetBridge[2412]: RTM_NEWROUTE: index:2
Aug 06 19:01:24 vmnet-natd[3207]: RTM_NEWADDR: index:2, addr:192.168.0.15
Aug 06 19:01:24 vmnet-natd[3207]: RTM_NEWROUTE: index:2
Aug 06 19:01:24 vmnet-natd[3207]: RTM_DELROUTE: index:2
Aug 06 19:01:24 vmnetBridge[2412]: RTM_DELROUTE: index:2
Aug 06 19:01:24 kernel: userif-3: sent link down event.
Aug 06 19:01:24 kernel: userif-3: sent link up event.
Aug 06 19:01:34 systemd[1]: NetworkManager-dispatcher.service: Succeeded.
-- Subject: Unit succeeded
-- Defined-By: systemd
-- Support: http://www.ubuntu.com/support
--
-- The unit NetworkManager-dispatcher.service has successfully entered the 'dead' state.

In the articles I've read about this, they all point to DHCP on the host being the culprit. Since I'm using a static IPv4 address (with IPv6 disabled), this should not be a consideration. Is there anything else I can look at? The network drop is killing my host for about 20-30 seconds - very annoying!

[RaSystemlord]

Don't know exactly since I have not seen this happening. I've used quite a bit NAT VMs in unison.

Cannot completely deduct the log, but are you saying that NAT is interrupting? I'm asking because if Bridged is the problem, it is completely another ball game.

As for the references to NAT ... don't know, but I found that there is something fishy with that in VMware. I've learned a couple of times, not sure if it was Ubuntu or Windows, that when using a static address, you STILL had to have DHCP active. It doesn't make sense, but it wouldn't work without DHCP being ON. Maybe that notion will help you, not sure?

[ajgringo619]

NAT is only being interrupted because my host network is resetting, otherwise it's been great. After testing my host with a static IP for the last 12-24 hours, the problem has been greatly reduced but not completely gone.

[RaSystemlord]

Do you mean that Host network itself keeps on resetting, like not getting into Internet?

If that is the reason, there are many reasons behind it, all of them Host OS-related, not VMware related. I'm not sure that this is what you are asking?

Well, anyway, if so, here is a list in some kind of likelihood order (not knowing anything about your network or OS or devices):

- router-box one port is failing. Change the port (if wired connection)

- router-box, actually a NAT-box, is in the end of life. They last a couple of years in the average, unless they are Cisco

- your Host network adapter is in the end life (over heating, perhaps)

- somebody else is competing with your wireless channel

- drivers on your Host are faulty (this really can interrupt the network in regular intervals)

- somebody has added a new DHCP-server ("accidentally", having no idea what he is doing) to your network, like from a mobile router (shouldn't really interrupt in regular intervals, just once in awhile)

- your wired cable is loose or under a bed/table stand

- ISP Name Server is not reliable. Should just block, not do it on regular intervals. Anyway, add the Google name server, 8.8.8.8 to make sure

All of these HAVE HAPPENED 🙂 ... symptoms can vary.

[ajgringo619]

While I will check all of your suggestions, I really don't think they apply here. According to links like these - https://fluentreports.com/blog/?p=717 and https://www.nikhef.nl/~janjust/vmnet/ - the problem is with VMware (and has been since WS version 8). I have not tried applying the patch (as I'm not using DHCP anymore). So far, the only times I've seen the userif-3: sent link down/up events has been at system boot and whenever I start a VM; both events are fine with me.

[ajgringo619]

FYI - this discussion should have been posted in Workstation, not Player. Hopefully it will get moved...

[RaSystemlord]

OK, your links explain your matter better - originally, Host OS was not mentioned. 

Not sure, if those links explain a real, generic, case, though. Why would DHCP-address be renewed in a live system? Never seen anything wrong in VMware, because of that (using VMware 15+years, in Ubuntu and Windows, lately in Kubuntu and Ubuntu Studio) in large corporate networks and elsewhere.

It seems like a faulty DHCP-server or its configuration. Often times, in corporate networks, you can ask for a long lease time (weeks) for an ip-address. That is often also a requirement, because of some licensing software. You may even ask for a DHCP-server never to change your ip-address. 

In your own networking, there is no need to use a DHCP-server, you can just as well use a static address, because networking is in your own control. The problem may arise when moving the computer between networks - which isn't really that common these days with limitations of all sorts.

Not sure, what this DHCP-server ip-renewal means in a live system? Normally, it only changes ip during the boot, IF it changes at all. If this is a specific case in a certain corporate network, THEN, I realize that you do have a problem. Sorry, I'm not a security-guy and do not know about all the specific networking arrangements that may be enforced for end users.

However, you say that the problem is also with a static address of the Host, which, if true, makes the writing above mute. Since no such phenomena are happening in a standard case, not in Kubuntu 20.04.2 or Ubuntu Studio 20.04 anyway (I have tested briefly many others), the problem is perhaps in something, which is not standard for Ubuntu, if that is what you have. I need also to add that I'm not using normally a static Host address. With normal, home-use, boxes (Zyxel, D-link, TP-link) I have never seen such a problem, in wireless or wired connections ... so, not a general case.

[ajgringo619]

As of now, the only time I'm getting these messages is when I start a VM; the userif up/down are still happening, but I'm not seeing a drop in the network on either the VM or my host PC (which should be the normal case. In both previous instances, I was getting notifications from NetworkManager about the disconnects (host and guests).

I really appreciate all of your responses. My only real concern going forward is why this is still being brought up after 8 complete revisions of the VMware client.

One more thing: in doing a simple search of kernel: userif-3, almost every system referenced in the links have VMware installed. In fact, after checking my system journal, I didn't get my first message until after I installed VMware.

[ajgringo619]

Interesting...now when I get the messages, I'm only losing the network on the guest, not the host. Since this seems to be tied into something that NetworkManager is doing in concert with VMware, I'm going to disable it for awhile and see if anything changes.

[RaSystemlord]

It's strange that the symptoms fluctuate. I don't think that this kind of situation is common and thus it feels like system specific (hardware or software).

As for my own experiences, there are reasons why they are perhaps not directly relevant to your case:

- with later versions of VMware, I'm using Player and not Workstation Pro. Since you mention kernel modules, I'm not altogether sure, if the situation is exactly the same with Pro and Player. Unfortunately, I'm not the right person to comment on possible problems with kernel modules.
- lately, I'm using Kubuntu. While the basis is perhaps the same as with Ubuntu, I think the management tools are different. Furthermore, default settings may be different

- I have also used Ubuntu Studio, which does have a different version of the kernel (low latency). I don't use (plain) Ubuntu these days.

As for networking settings, one thing came to my mind:
- network interrupts can happen, but if they are very short, you might not notice them. However, if you have Automatic Connection ON for several networks and IF it immediately tries to switch to another network, which is bound to fail, you may experience a problem. Never seen this happening, but I guess in theory this could be possible ... if the switching is too quick and several networks are set to automatically connect and switch. I have no direct idea, why this would be your case.

Exact Host OS version, network type and Guest OS versions, might ring a bell to somebody.

[ajgringo619]

Strange...I posted a workaround I found last night, but it's not here! I got rid of NetworkManager and the problem seems to be gone; not an actual solution, but a solid fix.

[RaSystemlord]

Glad that it works now!

Since Network Manager is a rather common name, which software you actually mean? Is it the delivered GUI for network management in Ubuntu 20.04 or what is it?

[ajgringo619]

It's the NetworkManager apps that are part of the standard install of Linux Mint 20.2 (Ubuntu 20.04):

$ apt search ^network-manager
p network-manager -
p network-manager-config-connectivity-debian -
p network-manager-config-connectivity-ubuntu -
p network-manager-dev -
p network-manager-fortisslvpn -
p network-manager-fortisslvpn-gnome -
p network-manager-gnome -
p network-manager-iodine -
p network-manager-iodine-gnome -
p network-manager-l2tp -
p network-manager-l2tp-gnome -
p network-manager-openconnect -
p network-manager-openconnect-gnome -
p network-manager-openvpn -
p network-manager-openvpn-gnome -
p network-manager-pptp -
p network-manager-pptp-gnome -
p network-manager-ssh -
p network-manager-ssh-gnome -
p network-manager-strongswan -
p network-manager-vpnc -
p network-manager-vpnc-gnome

Purged and blacklisted.

[RaSystemlord]

Thanks!

I don't know what difference Mint does with this, other than those possible differences mentioned already above. I installed, into VM, Mint to see how it looks. It looks, in this regard, different to Ubuntu and Kubuntu.

However, GUI probably makes no difference in this. The Debian packages (or non-Debian) that actually make the Network Manager, might make a difference (still depending on kernel and its modules, perhaps). Well, I'm not the right guy to hunt that down - perhaps you already started that, if it really interests you that much.

There are quite a few inconsistencies and failures with VMware Linux versions (*buntu, perhaps some of them are generic to Linux). For instance copying files between VM and Host is a big mess and results in tens of modal error messages if you try. (Of course, networking works for file sharing). I'm not saying that there aren't problems with Windows 10 versions.

It is only so much, ANYBODY on this Forum is capable of doing or mostly even willing to do without any knowledge on the internals of the software. It seems that software quality control (like certification, testing, releasing, documentation) is under staffed. Ubuntu versions used to be error-free in the past.

Idea of a Forum is not really to help people in their difficult problems. The idea is to prevent people struggling with problems by giving solutions beforehand. Now we are hitting a brick-wall in many cases.

[888BBQ888]

thx! another solution is to patch userif.c module https://communities.vmware.com/t5/VMware-Workstation-Pro/Vmware-disconnected-and-reconnects-frequent...

[Rondo85]

I am wondering, why VMware is not taking up this issue. Since Ubuntus 22.04 VMware had become more or less useless for me. This Issue is one of most hurting ones!

[vmwareh8r]

I am wondering this as well.  It seems like I'm always having compile modules by hand and do all sorts of other things just to get vmware to work.  It's incredibly annoying.