Mavericks Dropping Network connection on test Mac Mini server

Not at work, but my media centre / home macmini 2011 i5 that drives my house has been dropping ethernet intermittently, and very frustratingly.

It's plugged direct into TP-Link home router running gargoyle-router.com. I thought I had failing hardware, but this makes me think otherwise. Watching with interest.

I too have the same issue - Mini (2013) connected to Vlan with wireless MacBook Pro Retina (late 2012) both latest Mavericks.
Connection keeps dropping - repeated messages on macbook screen about lack of connection. Use Finder - reconnect and all is well for a while then it disconnects and messages start appearing again.
I have been having trouble with my MacBook using a TimeMachine disk on the server - I suspect it is the dropouts that is the issue with that too. Large transfers to the server disks take forever and only finish if I keep the connection alive.

the wifi issue could be if you are using bluetooth keyboard and mouse at the same time, there is a bandwidth issue where bluetooth devices will steal bandwidth from the wireless portion of the chip causing the wifi network to go up and down, or switch G band instead of N band. Apple should choose a better wifi chipset to address this issue, happens on a bunch of cheap windows laptops too.

It's too early to say for sure, but I'm not currently seeing this issue after upgrading to 10.9.4. Anyone else?

i can confirm your expectations.... im not seeing the issue either anymore...

over 10000 ping without a drop... before i use to have about 10% drop...

anyone else ?

So far so good on 10.9.4 here, but it's only been a few hours since I've upgraded our 2 problematic minis.

update on 10.9.4 !!!

62000 ping... no drop !!!!! think Apple finally fixed that one...

any updates from everyone ?

Issue returned for us after about 2 weeks on 10.9.4...I have a case open with Apple. The USB to Ethernet adapters do seem to resolve the issue though, but it needs to be fixed by Apple.

I've been watching my 18 Minis since the 10.9.4 upgrade. so far none have dropped their network connection. I really hope this issue wasn't just delayed with the 10.9.4 install.

I hope we are the exception and it does indeed stay resolved for you. The response I received from Apple was less than adequate...basically they said that the fix is scheduled for release with Yosemite and kept urging me to try the Developer Release to see if it resolved the issue(it seemed to fix it). IMHO, that is completely irrelevant to our current situation though. I have 172 machines that are currently running Mavericks(iMacs and MBP) that are affected by this issue. The way I see it is that the "feature" wasn't present in 10.8.x, they added it into 10.9.x, then fixed it in 10.10 so either back port the fix to 10.9 or remove it completely as in 10.8. I've also spoken with Cisco regarding this and Cisco's response was "that Cisco devices will NOT reply to UNICAST ARP REQUESTS and we will not support this in the future either."

We are still noticing this issue on 2 of our 10.9.4 mini's.

We suddenly started experiencing this problem on our two Mac Mini Servers (10.9.3 and 10.9.4) last week.
Some time ago we also had the Unicast ARP problem to our CISCO switches, which was resolved with net.link.ether.inet.arp_unicast_lim: 5 -> 0

Using a USB Ethernet adapter stopped the ping loss, but does restrict the connection to 100Mb. A Thunderbolt Ethernet adapter did not fix the problem.

For a permanent fix we did this:
• In System Preferences select your Ethernet connection that has ping errors and then choose 'Duplicate Service'
• Check the duplicate is a lower priority in 'Service Order' • Change the IP address on the duplicate to anything else. (I just selected DHCP)

I would do the above when at the console as the first time I did it when I change the IP address on the duplicate the server stopped answering to pings completely so consequently also lost RDP connection. Second time I did it form the console, deleted the duplicate, did it again with it working perfectly. No idea why it works, but we have since had 0% packet loss.

Credit to Ben Bass : https://groups.google.com/forum/#!topic/macenterprise/3gEKaZ7SBLk

I think this issue is very simple. The following was very helpful:
Posted 4/1/14 at 8:46 AM by tanderson
I just started seeing this late last week on my 10.9 Mac mini server. It had been running fine for weeks and just started up. I see network disconnect errors in the logs and have noticed that user sessions aren't getting disconnected when it happens (they hang around in the Connected Users tab). Threads on Apple support discussions about it too. Upgraded to 10.9.2 and no change.

I discovered this with a simple file share on a network application. App was continually getting a license error on one machine for a specific network application (too many user error or license already used). Server was OS 10.9.4 and the machine was OS 10.9.4. User was a laptop and she would pick up and take laptop home each night. By doing this, the server was disconnected but the server did not drop the connection. She would come in the morning, connect to server again, and the "Server" app would list her name twice under the "Connected Users" tab of File Sharing. Manually disconnecting the phantom user (the one that was idle for almost 20 hours) solved the problem. However, any network drop (forced or unforced) seems to cause this exact issue.

Very buggy stuff from Apple.

In reading other posts, I forgot to check the network adapter but I am assuming it is thunderbolt. Would be interesting to test with simple disconnects to on board Ethernet versus an adaptor.

You may want to enable AFP Idle Disconnect. I believe it is turned off by default to allow for AFP Network Shares to reconnect automagically.

We made the following changes:
sudo serveradmin settings afp:idleDisconnectOnOff=1
sudo serveradmin settings afp:idleDisconnectTime=120

Hey just to let you know that we have been tracking this for about 9 months... a full write-up is here on our blog here: http://www.macstadium.com/blog/osx-10-9-mavericks-bugs/

A couple days ago, one of our subscribers who happens to also be an Apple employee put us directly in touch with the CoreOS at Apple developers in Cupertino, CA. This was really exciting news because we have been trying to get an escalation at Apple via the support and beta communities for a very, very long time with no traction.

Apple has since validated the issue via a lab we setup for them here at Macstadium.com, and they are now looking at a driver stack problem being the main issue. The biggest clue was the Apple USB <> Ethernet adapter not presenting the problem on the exact same host as the internal Gigabit Ethernet adapter - on the same network - at the same time.

They are moving very quickly, so we are very optimistic for a patch for 10.9.x as well as the solution being rolled into the next 10.10 beta release. We'll keep everybody posted in coming days.

Not much to offer other than a "me too."

rseward's NIC duplication trick has so far worked for me after applying it today.

I thought it was an issue with the Netgear switches so I put in some loaner Cisco Sg-300 switches I keep on hand. That worked for about 5-6 days. But today it started again.

Macstadium,

I too have a case open with Apple regarding this issue and asked about the validity of the statement you made and this is the response I received:

"Hello Jerry.

Good morning. Just wanted to touch base and provide you an update to the information you provided. There has not been any validation to this being implemented into 10.9.x and or any updates that it would be. However, we do know that this is fixed in a future release, which you have already tested and validated. What I can do at this time is mark the case pending Future Release and if any updates that come up that I can share with you I will do so promptly at that time."

I was just wondering if you had any update from your end as this could simply be a case of right hand/left hand. The mentioned fix that is mentioned above is regarding Yosemite, but this is not a feasible option for us and I'm sure quite a few others. Thanks in advance.

I had similar issues with my MacBook Pro (late 2013) and explored a lot into this problem. Here[0] is a detailed explanation and solution.
http://pankajmalhotra.com/ARP-and-ethernet-issues-with-osx-mavericks/

Hope you find it useful.

Cheers !
Bitgeeky

Running this script from the MacMiniVault folks completely resolved the issue on a 10.9.4/Casper Suite 9.32 install at a school district I've been working at:

https://github.com/MacMiniVault/Mac-Scripts/blob/master/unicastarp/unicastarp-README.md

RobertHammen, do you know how would this would be reversible should things hit the fan? Or if-when Apple released their fix? I want to try this ASAP on one system.

Thanks!

Edit: Wow they wrote that script back on November 8. And Apple still hasn't released the patch.

gskibum, reversing the script would be removing the line from /etc/sysctl.conf and rebooting.

Ahh I see that script creates the file in the first place. I suppose besides editing the file it could be removed altogether?

You are correct, the file is only a container for commands so presence makes no difference.

I bought a USB3 to Giga Ethernet Adapter from Cable Matters ($20.00 Amazon) that fixed the problem using AX88179 driver. All Macs with Mavericks 10.9 to 10.9.4 have the same problem at work, with Wireless, Cable, USB to Ethernet, Thunderbolt to Ethernet. Have contacted Apple Care with no luck

It seems to be that it is an embedded hardware issue that is random thru each unit produced. A necessary evil. But embedding always causes some type of lobotomy somewhere if it isn't mapped out correctly.