SOLVED All services reset during SMB transfer

[tyler.montney]

Box has no Internet connection, so I rely on the CMOS clock. I realized it was off, so I corrected it. Running "date" from the shell shows an inconsistency.

• System > General > Time Format shows the correct time
• date shows a date that's 5 hours and 18 seconds behind.
  • Time zone is set correctly, even toggled it back and forth.

Not sure if an incorrect date/time would cause issues, but it certainly needs to be corrected.

 

[Redcoat]

Intel 82579LM.

My two TrueNAS X9-SCM-F boxes each have an LM and an L - the L's are not used, The LM's are on the 1G network. The boxes have a Solarflare 10G direct connection for replication of snapshots (one's a backup). My W10 box is always on the wire, the other two mostly on wireless.

Last evening my wired workstation Veeam backup reported moving 45.1 GB at an average 23.5 MB/s.

I'm out of opportunity to help on your issue, I suspect - but I will keep looking on. Good luck to get a solution.

 

[tyler.montney]

My two TrueNAS X9-SCM-F boxes each have an LM and an L - the L's are not used, The LM's are on the 1G network. The boxes have a Solarflare 10G direct connection for replication of snapshots (one's a backup). My W10 box is always on the wire, the other two mostly on wireless.

Last evening my wired workstation Veeam backup reported moving 45.1 GB at an average 23.5 MB/s.

I'm out of opportunity to help on your issue, I suspect - but I will keep looking on. Good luck to get a solution.

Appreciate you taking the time to help.

 

[tyler.montney]

I thought perhaps the NIC was faulty, so I threw in a TG-3468 I had lying around. I really thought that was it, but the same issue still happens. Bummer.

 

[tyler.montney]

I just now realized my main box is on U3 and the problem box is U1. I set up my first TNAS box not more than a few weeks ago and swear I pulled the latest ISO from the site. Didn't think that there were updates since.

Manually upgraded to U3, no change. Gonna try a clean install, can't think of anything else to do.

Edit: No change on clean install.

 

[tyler.montney]

Issue still persists with NFS, although it hasn't timed out yet. I'm running this off an old 2.5" HDD, perhaps that's the issue (there were some tests that said its SMART was failing; TrueNAS short test/SMART says it's OK). Although, I've read with TrueNAS caches heavily and most of it lives in memory. I only used a regular HDD because I had nothing else available.

Ordered an NVMe PCIe adapter and will try with a different disk.

 

[tyler.montney]

While I wait for the adapter, I tried a different HDD. Same result.

SSH is unaffected. Transfered x3 1 GB test files to /root and another set to the pool. All disks pass a short test.

 

[tyler.montney]

At a loss, opened bug report here: https://jira.ixsystems.com/browse/NAS-110495

 

[tyler.montney]

Turns out my box doesn't support NVMe as a boot drive, not without injecting a module into the BIOS. Since the main tool seems to be a leak and the other is third-party, I deem it too risky. Will settle for a SATA SSD.

 

[tyler.montney]

• Ran built-in memory diagnostics: Passed
• Upgraded BIOS to latest
• Upgraded TrueNAS to 12.0-U3.1

Issue still persists.

 

[tyler.montney]

Figured I'd do a sanity check and use Veeam on my main box (12.0-U3). Works perfectly.

 

[tyler.montney]

I got my SSD and did another go with transfers, same issue. However, this time I was connecting from my isolated backup network. During the transfer, my RDP session got interrupted to the Veeam box. "Strange" I thought. Then it occurred to me that, the issue may be with the Veeam box and not with TrueNAS (since you know I changed just about every thing I could on it).

Turns out, I had put a PCI-E card in a PCI slot. Not that I ever thought that would work, I guess I assumed it was PCI-E. All my tests were done from the other network, which was on a properly installed NIC. My SSH transfers were direct, not through the Veeam box. Somehow, despite all my thorough testing, I managed to not come across this until now.

Long story short, PCI-E in a PCI slot is bad news. It'll work, but it won't take the strain.

 

[Redcoat]

Ah-ha!! The Eureka moment.
Glad you found it finally.