ethernet connection fail between two mac

[cebroth]

Hello Everyone

Hope someone can help me with an annoyng issue:
I have a small office of video editing with my collegue: we have a Mac Pro late 2013 and a iMac 5K retina and we use a MacPro 1,1 as server for data storage. the connection is via ethernet and we can work perfectly in full hd and 4k etc with the data contained in our "server". the weird problem comes as we transfer a queue of video files from our working stations to the server. for example: i connect my sd card to the Imac and I want to trasfer the data (let'say 20 GB of videos) to the server: at a certain point the connection fails an the process is interrupted. the same happens (this is worse) when in Premiere/media encoder i start a queue for creating proxies: the process starts and after a while the server disconnect, the icon disappear, the encoder freezes and I have to restart the mac-pro-server.
I thougt it was an ethernet swithc issue and I try to connect directly one workstation to the mac-pro-server, same problem: even with a direct connection: the connection fails and everything freezes. Anyone have any idea? thanks in advance

 

[Ruggy]

Hello Everyone

Hope someone can help me with an annoyng issue:
I have a small office of video editing with my collegue: we have a Mac Pro late 2013 and a iMac 5K retina and we use a MacPro 1,1 as server for data storage. the connection is via ethernet and we can work perfectly in full hd and 4k etc with the data contained in our "server". the weird problem comes as we transfer a queue of video files from our working stations to the server. for example: i connect my sd card to the Imac and I want to trasfer the data (let'say 20 GB of videos) to the server: at a certain point the connection fails an the process is interrupted. the same happens (this is worse) when in Premiere/media encoder i start a queue for creating proxies: the process starts and after a while the server disconnect, the icon disappear, the encoder freezes and I have to restart the mac-pro-server.
I thougt it was an ethernet swithc issue and I try to connect directly one workstation to the mac-pro-server, same problem: even with a direct connection: the connection fails and everything freezes. Anyone have any idea? thanks in advance

The only thing I can think of is that possibly you've got an energy setting which tells the disk to go to sleep after a specific time like 15 minutes.
I have had problems with that on transfer as for some reason it doesn't pick up that it's working and powers down. Used to happen to me a lot when downloads on dial up could take hours.
Sorry if that doesn't help, it's all I can think of.
Good luck

 

[hobowankenobi]

Can be tuff to diagnose. As mentioned, need to verify nothing is sleeping (OS or drives) via energy settings. Also verify that neither machine is set to log out automatically after inactivity (Security settings).

If that is not the issue, next would be reviewing all network connections: cables, termination, any switches involved.

Does the connection stay stable for long periods (many hours, days) with no data moving?

There is an outside chance that there is a disk I/O error, lag, stutter, etc. If that happens, it could interrupt the connection long enough to generate an error or disconnect.

AFP or SMB? Might try changing to see if it behaves any different.

You can look at some networking tools to look for any dropped packets or throughput issues.

 

[cebroth]

Hi. thanks for the replies. I don't think is a energy setting because I set hard disks sleeping function on "never".

for hobowankenobi:
about the network connections (cables, termination, any switches): the problem occours even if the link between the computers is direct (ethernet port to ethernet port).
the connection, except during the transfert of multiple video files or the queue for generating proxies or conversions, is stable and usually for a day and more, working or not (let say if we are out of office for 2/3 hours or more): the problem occurs just when there is a batch writing process on the hard disk... so I don't think can be a disk error.
AFB or SMB -> how can I try?
do you have any suggestions about a networking tool software that I can use?
thank you very much

 

[hobowankenobi]

the problem occurs just when there is a batch writing process on the hard disk... so I don't think can be a disk error.

If the error only occurs under heavy load...sounds like it is a disk or IO problem...or a network bobble that is causing the writes to pause.

As for network, how do you connect?

afp://
smb://

 

[cebroth]

thanks and scuse me for the dealy: I tried both with afp:// and smb:// and the problem occours the same way as usual. today I got the disconnection just moving a 1GB file from HD1 to HD2 of the Macpro server managing the process from one of the workstations. I mean that from the workstation I just moved a file between two hard disks of the server... suddently the connection interrupted. this is a weird new entry....

 

[hobowankenobi]

Strange. Still, good to narrow down variables:

• Not specific to AFP or SMB (both behave the same)
  
• Not a router or switch or other network backbone (all have been removed)

So that narrows it down to the most likely causes:

• Network hardware (one of the NICs in either box, or perhaps the connecting cable)
• Network software (something amiss at a low level in one of the OSes regarding network config/hardware drivers)
• Storage hardware (one of the storage devices, a cable, bus or other related piece of hardware)

Removing the NIC/OS variables:

If you had a Thunderbolt or USB 3 external NIC, you could try that and see if anything changes. On Mac Pros, most have two NICs...could try the second one too, although they are likely exactly the same chipset and a shared board. Have you verified that you have a gigabit connection? that would show that all 8 contacts are engaged and working. You can still use Ethernet with as few as 4 of the 8 wires, but at slower speeds. If you can get a stable (never disconnects) mounted share, that would point to a NIC or OS drive issue.

Removing hardware storage variables:

One could try using an external device, in order to isolate the internal drive and bus. Same for both the send and receive boxes. If possible, read from and write to externals; if no issues and no disconnects (keeping in mind performance may be reduced depending on the storage and bus connection speed) that would point to internal storage. Then next step would be trying to isolate if it is a read issue, or a write issue.

Some or perhaps all of this could be done through scouring through logs, looking for dropped packets, read/write errors, etc. I tend to swap out hardware to try and find the problem....but that is only handy if you have spare hardware resources at your fingertips. Some tools can do the testing and alert to a problem might be useful too. One that has been around forever that is good to test performance on mounted volumes is Helios LanTest. It will test all sorts of reads and writes, and lets you loop many tests automatically.
[doublepost=1552936244][/doublepost]Reread your latest post. Leaving everything above, but If you have a issue between two internal HDs....you have narrowed down to:

1. One (or both) of the HDs
2. The SATA ports the drives are pluggged into
   
3. The SATA bus they connect to (logic board)
4. Outside chance there is an OS issue

If you have open HD slots, you might try moving drives to the other/empty bays. Preferably one at a time, and repeat test.

If you have bad ports, or even the entire SATA bus (that could cause issues on all internal SATA ports) you could consider a supported third party PCIe SATA controller, bypassing the integrated (on the logic board) SATA controller. Something like this. Would have to get creative with SATA cables. Perhaps there is a better option with at least 2 internal ports...

 

[cebroth]

Hi. I did a lan test with lantest but everythinglooks ok. Today i tried moving files (giving the commands from the two workstations) between the two internal drives of themac pro server. I tried from the two workstations, changing the protocols, from hd 1 to hd2 and reverse, with only one workstation connected an with both workstations connected.weird results: moving from the firts workstaton (mac5k) every attempt was successful, from THE MACPRO there was same fail that suddently stopped the connection. So i tried again for verifing: i use the same settings but this time the requested file transfercompleted successfully.... looks very random. tomorrow iwill try with some file swap between external HD connected via usb or thunderbolt trying to find a rule in this problem. Btw after every connection crash the connection to the server is impossible from every workstation and the server need to be restarted. Thanks for your help in this jungle

 

[hobowankenobi]

Hi. I did a lan test with lantest but everythinglooks ok. Today i tried moving files (giving the commands from the two workstations) between the two internal drives of themac pro server. I tried from the two workstations, changing the protocols, from hd 1 to hd2 and reverse, with only one workstation connected an with both workstations connected.weird results: moving from the firts workstaton (mac5k) every attempt was successful, from THE MACPRO there was same fail that suddently stopped the connection. So i tried again for verifing: i use the same settings but this time the requested file transfercompleted successfully.... looks very random. tomorrow iwill try with some file swap between external HD connected via usb or thunderbolt trying to find a rule in this problem. Btw after every connection crash the connection to the server is impossible from every workstation and the server need to be restarted. Thanks for your help in this jungle

So that seems to narrow down the the server is the source of the issue...and that a reboot is needed to restart file services or some other network related process.

If this happens while monitoring, note the time of the disconnect event and then check out the console log about the same time for clues.

Try the second NIC in the server? If it still does it...it rules out the port itself. Still leaves the OS (or some software issue), and the network chipset on the logic board as likely culprits.

Anything installed or running on the Server itself? I wonder if you can do file sharing booted to Safe Mode....as a test? Worth trying.

Another option, that will take more time: if you have a free internal bay on the server, a freshly installed OS on a newish (known good) HD or SSD might be worth testing, though it may not make it easy to know if it is OS or drive....but both are easy to replace, so why not?

What OS is on the server now? Might be worth exploring options for a different flavor is none of the above sheds new light.

 

[cebroth]

Hi. thanks.
before I go on with the problem research I want to ask you a (maybe stupid) question. I never report here that the computer we are using as a server (Mac pro 1.1 - 10.11.5 El Capitan) is a normal machine without any server manager installed: I mean: is a normal operative computer with the OS and the various applications installed. Now I have the doubt that we need something like "mac OS server" installed. In your opinion is an explication for this problems I have? in this case is possible to install the server manager side by side to the OS or I necessarly need to reset the machine form making a server-only computer out of it?