apx

Could you drop what's your SMB config in OS. (powershell - Get-SmbServerConfiguration) Mine:

On my end, if these bad apple file are inside another archive or other file this also breaks transfers. In this long ongoing topic, people report that bonding network cards may help. When I tried to tinker with network adapter settings, I only managed to bork entire config, and had to reset) Plus it's already sounds strange, why 2 network adapters would affect certain file transfer. https://forums.unraid.net/topic/79454-file-transfer-results-in-error-0x8007003b/ @JorgeB Whats your environment, Windows 11 ?

True, but would be great to figure out the reason. Because such files bork also network backups as mentioned in initial post. So you run backup, and it fails due to certain files. Had encountered same issue when made Game folder backup. Still would try to gather more evidence. Maybe more samples can help.

Unfortunately - same story.

Here are samples (https://gofile.io/d/ynZJ8m). Maybe someone else can reproduce it on their end, and see what's wrong with the files. TLDR - original files FAIL to copy. (rarely it goes through) (Store method) - repacked files FAIL if STORE compression method is used. - repacked file PASSes if DEFLATE method is used.

Heard this opinion too, tried to disable it, but same story. It's just 2 ZIP files from thousands similar. One had whitespaces in archive content filenames, but doubt it had something to do with it. With expanded SMB logs, process has these events. Those 65536 values doesn't look right, like it could be some kind of limit. May 24 13:10:14 SHAPXUB smbd[12701]: [2026/05/24 13:10:14.900685, 3] ../../source3/smbd/smb2_write.c:214(smb2_write_complete_internal) May 24 13:10:14 SHAPXUB smbd[12701]: smb2: fnum 902811720, file New Folder/2 - fail.zip, length=65536 offset=11534336 wrote=65536 May 24 13:10:14 SHAPXUB smbd[12701]: [2026/05/24 13:10:14.900927, 3] ../../source3/smbd/smb2_write.c:214(smb2_write_complete_internal) May 24 13:10:14 SHAPXUB smbd[12701]: smb2: fnum 902811720, file New Folder/2 - fail.zip, length=65536 offset=10027008 wrote=65536 May 24 13:10:14 SHAPXUB smbd[12701]: [2026/05/24 13:10:14.901193, 3] ../../source3/smbd/smb2_write.c:214(smb2_write_complete_internal) May 24 13:10:14 SHAPXUB smbd[12701]: smb2: fnum 902811720, file New Folder/2 - fail.zip, length=65536 offset=11599872 wrote=65536 May 24 13:10:14 SHAPXUB smbd[12701]: [2026/05/24 13:10:14.901430, 3] ../../source3/smbd/smb2_write.c:214(smb2_write_complete_internal) May 24 13:10:14 SHAPXUB smbd[12701]: smb2: fnum 902811720, file New Folder/2 - fail.zip, length=65536 offset=10485760 wrote=65536 May 24 13:10:14 SHAPXUB smbd[12701]: [2026/05/24 13:10:14.901713, 3] ../../source3/smbd/smb2_write.c:214(smb2_write_complete_internal) May 24 13:10:14 SHAPXUB smbd[12701]: smb2: fnum 902811720, file New Folder/2 - fail.zip, length=65536 offset=10092544 wrote=65536 May 24 13:10:14 SHAPXUB smbd[12701]: [2026/05/24 13:10:14.901926, 3] ../../source3/smbd/smb2_write.c:214(smb2_write_complete_internal) May 24 13:10:14 SHAPXUB smbd[12701]: smb2: fnum 902811720, file New Folder/2 - fail.zip, length=65536 offset=11665408 wrote=65536 May 24 13:10:14 SHAPXUB smbd[12701]: [2026/05/24 13:10:14.902185, 3] ../../source3/smbd/smb2_write.c:214(smb2_write_complete_internal) May 24 13:10:14 SHAPXUB smbd[12701]: smb2: fnum 902811720, file New Folder/2 - fail.zip, length=65536 offset=10551296 wrote=65536

Here is uncertain thing. I managed to narrow it down to particular ZIP files. (like similar was mentioned also for TrueNas - https://www.truenas.com/community/threads/copying-odd-zip-or-msi-file-causing-smb-service-to-freeze.87066/ ) (if those are repacked to 7Z it's all good and smooth) Like if such files is copied to NAS, transfer hangs a bit and becomes unresponsive. What would be proper way to pinpoint why system hangs during such file process? SMB log level, or there are other ways which may help?

Strange discovery, when backing up files with Macrium from W10 machine to Unraid, noticed that tasks fail on certain files/folders. And those even fail if I move files/rename long path to smaller one/rename end ZIPs to simple numbers. More confusion adds fact that 1000+ zips - FAILS, and 1000 is fine. Like literally those 100 are bad apples. (Okay 1000+1000 also fails) Even if I ZIP all those 1100 into single one, copying this to NAS fails as well. Zipped 1000 into one piece also goes smooth. Whaaaa. Could this be an issue because NAS runs with 4GB of RAM? While dashboard reports its barely utilized, otherwise I cannot come with anything why this strange scenarios may happens.

From few more tests, it seems Macrium Reflect doesn't handle those long file transfers straight to network disk. While VEEAM Agent handles this better. Also have seen posts where Macrium users mentioned that they do backups usually firstly disk to disk, and later move those images with more robust and reliable copy methods.

Probably will try to crank power mode to Best Performance, and see if it's better compared to Balanced. Could be also a thing when one SMB transfer is going, and some app tries to check network location, maybe in certain conditions it may have some impact.

I had SAMBA logs enabled previously, and monitored whole transfer. And those adding interface messages lead me to think, could this be interface bonding/bridging issue? I guess if interface were dropping, this would be obvious in logs.

Here are logs just after event. I start to see pattern, with 1 hour and 10 minutes. shapxub-diagnostics-20260403-1334.zip

I'm suspecting that my NIC in NAS goes down and re-appears. Like there are ton of write event, and whoops network interface appeared. New diagnostics added. After event occurred. diagnostics-20260402-2229.zip

Literally just now.

Its not like first time I have noticed, that huge file copying over network can result in errors. From what I have seen during file copy sometimes, speed goes to zero, and then it comes to software can it handle it and wait to retry, or fail. Here example case. Local Windows 10 machine file backup to network share, via MacriumReflect. Apr 2 20:12:41 SHAPXUB smbd[7525]: [2026/04/02 20:12:41.202711, 0] ../../source3/smbd/durable.c:928(vfs_default_durable_reconnect) Apr 2 20:12:41 SHAPXUB smbd[7525]: vfs_default_durable_reconnect: default_durable_reconnect_fn [GAMES--2026-04-02T18.33.49--00-00.mrbak.tmp] failed: NT_STATUS_OBJECT_NAME_NOT_FOUND durable reconnect code is shown here https://github.com/avati/samba/blob/e24fe5705e3c4d33705ebb584ea2009bb4a1a82c/source3/modules/vfs_default.c#L2429 I'm here more for advice, where to start investigation. Which kind of settings to check. And it fails in Macrium when task processes these files. https://forums.unraid.net/topic/198087-smb-drops-connections-on-long-file-transfers-and-on-certain-zip-files/#findComment-1624535 diagnostics-20260402-2054.zip

Then probably I made fine choice. Those huge files are mainly from backups.

Yea, there are even TB big backup files. Will stick for those cases - direct disk paths.

TLDR. If cache is 240, and I have single files bigger than that. What minimum free space should be set, so smaller files go onto SSD, and exceeding ones - skip cache. Shall I set there 200GB? Or plain simple, those huge copies should be done directly to disk, so it avoid cache entirely ? (I like this option more)

Will leave disk shares available, just hidden. Because for example cache is 240GB, and if I push 1TB single file to user share which is has CACHE->ARRAY setup, it starts to place file into cache. I dont think it will end well. ChatGPT said that it will avoid cache if file exceeds cache, but reality shows different story. Not first time when I doubt AI responses, and those are actually wrong.

Yea, I did configured like this, here I asked if I'm on right path or this has drawbacks and maybe better ways there are.

Hello, Currently my setup has 3 drives in Array, and ideally I want to control what goes to what drive. At first I had disk shares enabled, and that works well. Now managed to add cache, and it seems that there is no way to move cache to disk shares. Am I right that in above setup more optimal way would be?: make user shares for each drive. configure shares to access particular disk. (Share1 -> Include DISK1, 2 -> 2,...) add cache mover for each, so data from cache is moved to according disk over time.

diagnostics-20260316-1827.zip Meaning there may be other reasons for such behavior?

Hello to this community, would like to ask for some advice/suggestion/idea. Setup. Plain simple, array is is few 20TB drives, no RAIDs, no parity, no cache... 1Gbit NIC. Copying from Windows machine (1Gbit NIC too) directly to \disk share speeds start at gigabit, but slowly drops to half. That's 10GB single file. I know that those are turtle speeds, but shouldn't drive sustain speeds at least around ~100MB/s? And what could cause those slowdowns? iperf tests shows that both machines can exchange at 950Mbits/sec. (should be at least around 100MB/s) same drives installed locally maintains speeds around 150MB/s, when same file is copied to it. SMB Multi Channel setting didn't changed much. Will add cache eventually, but yet not sure if same wont happen with even simplest SATA SSD. Solved. Settings > Power Mode > Best Performance or Balanced.