[Solved] Packet loss across multiple NICs

[Skleman]

Deployment:
10G x520 NIC. Latest (4485) release. I am getting roughly 30% packet loss. I replaced the card, the transceiver, the fiber cable, the transceiver on the switch and moved the port with the same result. I can also ping the interface IP and localhost without loss. What's super weird is that I could achieve 300M out of it in this state when pulling multiple streams to backup through NFS mounted on another machine. Nonetheless, packet loss on ping. I then moved it to 1G thinking it may be a driver issue. Initially, everything was flawless but after about a day, I'm experiencing the same thing on that link.

None of the other machines in the network are experiencing anything of the sort and it's a home network (overkill, I know) so I'm not hammering it very hard.

ZFS with raidz2, hot spare (I know, should've done z3 but i'll get to that another day). Cache is on an SSD and another SSD for swap. I've got 64G of ram and not doing any dedupe or compression.

One thing I JUST noticed is that I'm having an issue where changes with mDLNA lock up the UI - another post on that - but the eureka moment was when I clicked disable on mDLNA (which locked up the UI), now I can ping continuously. I don't know if the symptoms are related but since I've exhausted every other google thread and everything I can think of, it may be something worth noting.

ANY suggestion is appreciated.

Thank you

[Skleman]

For anyone interested, I believe I've found the culprit. It would seem that any time the mDLNA service is running, I lose 30% of packets. I will be moving the server back to 10G to validate and will update the thread accordingly. I have no idea why DLNA would cause this but it's reproducible so I'm going to move the service to another server and disable it on NAS4FREE to keep the storage connection stable.

[raulfg3]

For anyone interested, I believe I've found the culprit. It would seem that any time the mDLNA service is running, I lose 30% of packets. I will be moving the server back to 10G to validate and will update the thread accordingly. I have no idea why DLNA would cause this but it's reproducible so I'm

Sorry but difficult to reproduce, because use of 10GB NIC for home use ( DLNA is for home use) is still not very common, but thanks to post, so other user can benefit from your experience.

[Skleman]

For anyone that comes across this, I have confirmed it is in the DLNA service causing my issues. After disabling it and working out the possibilities, DLNA causes the drops. Now, in addition, I've attempted to run DLNA on my Ubuntu box pulling inventory from NFS mounts off N4F. The same thing happened there. I suspect, but have no understanding to dig further, that something in my network doesn't like the DLNA broadcasts. I'm using an HP procurve 6600 and again, I suspect it's something with that as the single hop to my switch has the drops. Marking this solved. It's not by any means a N4F issue.

Thanks everyone