[/mnt/user dissappeared]

21 minutes ago, meebles6767 said:

<snip>

snips are meaningless.  diagnostics.zip has meaning.

[[6.6.1] AMD Reboot bug]

8 hours ago, J89eu said:

Let's hope 4.19 kernel comes soon

Checked 4.18.11 and above mentioned patch is not in that release.

 

Illustrates the two-edge sword: we have to keep up with kernel releases in order to get the latest device support and other device-related fixes; however, it also means we get the latest regressions introduced into the kernel which no one has previously reported.  That's the price we pay...

[6.6.1 Reboot Bug]

We change EFI- to EFI and/or in USB Flash Creator tool have the UEFI checkbox checked by default.

 

The reason it's not enabled by default now is that there used to be quite a number of buggy UEFI/USB boot motherboards out there and Legacy always just worked.  We added UEFI boot initially because some motherboards required it for proper VM functionality.

[Mover issues ***Solved***]

Changed Status to Closed

Changed Priority to Other

[SSH log spam during rsync]

Changed Status to Open

Changed Priority to Annoyance

[SSH log spam during rsync]

I've noticed it too, that ssh has become more "chatty".  Briefly looked into it but didn't see any obvious config file changes, but yes this is annoying.

[6.6.1 Reboot Bug]

2 hours ago, Hoopster said:

Kernel and other changes in later versions of Linux mean that many motherboards no longer boot into unRAID properly in legacy mode, but UEFI works.

Thank you for the help in this topic.  What do you think, maybe it's time that we make UEFI boot the default now?

[Bridge 0 Seems to be gone with 6.6]

Changed Status to Closed

[unRAID 6.6.x SMB/NFS access issues]

The issue is that FUSE crashes which causes 'shfs' to crash which causes 'nfsd' to crash.

 

Here is the code in FUSE (fuse.c) where the crash is occurring:

static struct node *get_node(struct fuse *f, fuse_ino_t nodeid)
{
        struct node *node = get_node_nocheck(f, nodeid);
        if (!node) {
                fprintf(stderr, "fuse internal error: node %llu not found\n",
                        (unsigned long long) nodeid);
                abort();
        }
        return node;
}

This function is being called with 'nodeid' 0, which is straight out of the NFS file handle.  That ‘abort()’ is what ultimately results in bringing down /mnt/user and along with it, nfsd.

 

This is almost certainly due to a kernel change somewhere between 4.14.49 and 4.18.  Trying to isolate ...

[unRAID 6.6.x SMB/NFS access issues]

It's all the same bug man.

[docker is unaccessible when losing Internet connectivity]

Changed Status to Closed

[docker is unaccessible when losing Internet connectivity]

4 hours ago, Soulrob said:

When I first contacted support, they thought I didn't have a license installed, apparently if you're using a trial license this is how unraid behaves - so I assumed it was an issue around licences.

Whoever told you that is wrong.  Please fwd response to me at [email protected]

Trial key works exactly like a Pro key except it does require an internet connection to validate, but only once following each server reboot.

[6.6.0 Realtek Nic Losing All Network Connectivity]

This might solve the problem:

https://www.spinics.net/lists/netdev/msg525992.html

 

However, I would like to wait for that patch to make it's way to 'upstream', could appear as soon as 4.18.11.  We'll have to watch for it.

[[6.6] Web Gui Fail to start after graceful reboot]

Changed Status to Closed

[/mnt/user dissappeared]

2 minutes ago, edgedog said:

I've tested 6.6.1 successfully for the past 3 hours with fuse_remember set to 0 without crashes of nfsd or shfs. Unfortunately my client experienced periodic stale file handle error messages. I also experienced these stale file handle error messages in 6.6.0.

I'm now going to test with fuse_remember set to 600 (a value which seems to allow my client to not experience stale file handle messages). I'll report back in several hours. Thanks!

Thanks for your report.  In practice you will probably not be able to run with NFS and fuse_remember set to 0 because there will inevitably be 'stale file handle' errors.

[Unraid 6.6.0 GUI Mode is Slooooow]

18 minutes ago, csmccarron said:

Just updated to 6.6.1, the issue is still present.

 

The GUI looks great by they way!

Thank you for the report.  I think this issue is related to another where for some users display is stuck at 800x600 - we haven't got a chance to look into this yet.

[[6.6.0-rc2] VM Memory allocation across numa boundaries]

FYI: The 'hwloc' package which includes 'lstopo' command is included in Unraid OS 6.6.1 but only available in Desktop GUI boot mode.

[[6.6.0] Docker CPU usage > 100%]

Changed Status to Retest

[[6.6.0] Kernel Panics]

Changed Status to Retest

[[6.6.0] NFS Kernel crash]

Changed Status to Retest

[[6.6.0] NFS Kernel crash]

Please retest with 6.6.1 since there is an update to FUSE which solves a crash, but I don't know if this is the root cause.

[/mnt/user dissappeared]

Changed Status to Retest

[/mnt/user dissappeared]

8 hours ago, edgedog said:

Is it possible to increase the logging or turn on debugging for the shfs process? It crashes right around the time nfsd does but I see no logs about why it crashed even though they're being piped to logger.

Yes, may come to that but log fills up fast.  Please retest with 6.6.1 since there is an update to FUSE which solves a crash, but I don't know if this is the root cause.

 

8 hours ago, edgedog said:

This wiki manual (https://wiki.unraid.net/UnRAID_Manual_6#NFS_.28Network_File_System.29) says that NFSv4 support has been included in unraid 6

The wiki is incorrect.  At this time Unraid OS only supports NFSv3.

[[6.5.3, 6.6.0-rc2] Duplicate MAC from different NICs (Aquantia & Intel)]

Changed Status to Closed

[[6.6.0] NFS Kernel crash]

One more thing to set up: boot the server in non-GUI mode, ie, Normal mode and have monitor hooked up (or use IPMI).  Then when crash happens see if any messages spit out there.