Interstellar

This started a while ago after a W11 update I assume, putting the VM to sleep and then waking it again results in 100% cpu time for the assigned cores and 40W+ CPU usage. Force Stopping and then starting again all is OK. However what is worse now is that seemingly the VM is starting up on its own accord and sitting at 100% CPU time for days on end. Anyone know if this is a common issue or not? If not I'll have to set some time aside to get diagnostics... nothing in the syslog at the time of the 100% CPU usage though. I can't just stop the VM as then the GPU doesn't go to sleep and my idle power doubles...

Bit perturbed by the 2GB USB stick issues, still using the USB stick I got when I started with UnRAID 5. Are we saying 4GB is the minimum now, if so, no mention in the release notes unless I’m being blind. Going to tidy the USB stick and try it anyway..!

Speedtest-Tracker no longer works for me.

Got an odd problem: If I don't set a --dns=10.10.2.1 flag in extra parameters, I can login to NPM fine from 10.10.1.X if I set --dns=10.10.2.1, then I get a bad gateway error on login as I'm accessing it from 10.10.1.X Problem is NPM is on a separate NIC, with the 10.10.2 range, and without setting the --dns=10.10.2 flag I don't get DNS resolution (opnsense + adguard). What is going on? 😶

I know this is the process, but posting that took a few mins just to see if it was just me or not, which has resulted in someone else having the issue so I know it’s not just me. sadly sanitising my diagnostics takes ages as it’s poor at anonymising everything. Its another thing I have to raise a bug for, but sadly my time is exceptionally limited at the moment.

Anyone else having issues where the UnRAID GUI stops working? The login page pops up, I enter user/pass as normal, click login then... nothing, just hangs in both Safari and Chrome. Tried /etc/rc.d/rc.nginx reload - no change. As far as I can tell the server itself is working normally, SSH works, dockers/VMs are all running. Absolutely nothing in the logs relating to nginx at all. Uptime was since a day or two after .10 was released and I'm using macvlan already. Anyone else?

Unsurprising that this is leading to some concern and misconceptions. Its good that us long time licenses are grandfathered but at the end of the day I paid $100 or whatever 10 years ago and have had many thousands of dollars of dev for that money. That sadly does not pay the bills at Limetech! With this new plan UnRAID keeps working albeit without updates, this means you haven’t lost functionality, you technically got what you paid for, which was the use of UnRAID in perpetuity and updates for a year. I’d imagine quite a few basic users don’t even remember to update that regularly anyway! Security is an interesting topic, not sure what side of the fence I sit on, if you want to be up to date then pay up or should security updates for an additional year be covered? From LT’s PoV, this can only be done by maintaining multiple branches and I don’t think LT is big enough for that, so I wouldn’t blame them for saying the former… It wasn’t so long ago that you’d pay regularly for new SW versions, I used to buy every new version of Lightroom… Long story short is that a balance needs to be struck and I don’t blame LT going down this route, it seems, a reasonable balance. If I was buying again I’d just get the top tier and be done with it.

Still looking for a way to manually start a verification process rather than temporarily setting a verification schedule and then once its started turning it off again....

Restarted... As requested: zfs list NAME USED AVAIL REFER MOUNTPOINT master 9.11T 3.48T 104K /mnt/master master/Mac HD 2 3.94T 3.48T 3.94T /mnt/master/Mac HD 2 master/Mac HD 3 5.09T 3.48T 5.09T /mnt/master/Mac HD 3 master/Working Files 76.8G 3.48T 76.8G /mnt/master/Working Files zfs mount master /mnt/master master/Mac HD 3 /mnt/master/Mac HD 3 master/Working Files /mnt/master/Working Files master/Mac HD 2 /mnt/master/Mac HD 2 showmount -e Export list for TOWER-NAS: /mnt/master <IP list> /mnt/cache <IP list> /mnt/disk1 <IP list> Trying to mount via the command that worked in 6.11.5: mount -t nfs TOWER-NAS.local:"/mnt/master/Mac HD 2" "/Volumes/Mac HD 2" rpc.mountd[12392]: authenticated mount request from IP:870 for /mnt/master/Mac HD 3 (/mnt/master) rpc.mountd[12392]: request to export directory /mnt/master/Mac HD 3 below nearest filesystem /mnt/master Master is setup as Private and: <IP>(sec=sys,rw,anongid=100,anonuid=99,all_squash) Thanks

Progress! Thanks - we're now mounted and working, but the startup of the array is so long that I'm getting GUI timeout errors.. Sep 14 12:37:02 TOWER emhttpd: /usr/sbin/zpool import -d /dev/sdh1 2>&1 Sep 14 12:37:03 TOWER emhttpd: pool: master Sep 14 12:37:03 TOWER emhttpd: id: 735173779397214344 Sep 14 12:37:03 TOWER emhttpd: shcmd (634497): /usr/sbin/zpool import -N -o autoexpand=on -d /dev/sdh1 735173779397214344 master Sep 14 12:37:05 TOWER emhttpd: /usr/sbin/zpool status -PL master 2>&1 Sep 14 12:37:05 TOWER emhttpd: pool: master Sep 14 12:37:05 TOWER emhttpd: state: ONLINE Sep 14 12:37:05 TOWER emhttpd: config: Sep 14 12:37:05 TOWER emhttpd: NAME STATE READ WRITE CKSUM Sep 14 12:37:05 TOWER emhttpd: master ONLINE 0 0 0 Sep 14 12:37:05 TOWER emhttpd: /dev/sdh1 ONLINE 0 0 0 Sep 14 12:37:05 TOWER emhttpd: errors: No known data errors Sep 14 12:37:05 TOWER emhttpd: shcmd (634498): /usr/sbin/zfs set mountpoint=/mnt/master master Sep 14 12:37:05 TOWER emhttpd: shcmd (634499): /usr/sbin/zfs set atime=off master Sep 14 12:37:05 TOWER emhttpd: shcmd (634500): /usr/sbin/zfs mount master Sep 14 12:39:01 TOWER emhttpd: shcmd (634505): /usr/sbin/zpool set autotrim=off master Sep 14 12:39:01 TOWER emhttpd: shcmd (634506): /usr/sbin/zfs set compression=off master Sep 14 12:39:01 TOWER emhttpd: shcmd (634507): /usr/sbin/zfs mount -a Sep 14 12:39:58 TOWER nginx: 2023/09/14 12:39:58 [error] 12456#12456: *255109 upstream timed out (110: Connection timed out) while reading upstream, client: 10.10.1.110, server: , request: "POST /update.htm HTTP/1.1", upstream: "http://unix:/var/run/emhttpd.socket:/update.htm", host: "TOWER.local", referrer: "http://TOWER.local/Main" Sep 14 12:40:57 TOWER emhttpd: shcmd (634512): sync Sep 14 12:40:57 TOWER emhttpd: shcmd (634513): mkdir /mnt/user0 Sep 14 12:40:57 TOWER emhttpd: shcmd (634514): /usr/local/bin/shfs /mnt/user0 -disks 14 -o default_permissions,allow_other,noatime Sep 14 12:40:57 TOWER shfs: FUSE library version 3.12.0 Sep 14 12:40:57 TOWER emhttpd: shcmd (634515): mkdir /mnt/user Sep 14 12:40:57 TOWER emhttpd: shcmd (634516): /usr/local/bin/shfs /mnt/user -disks 15 -o default_permissions,allow_other,noatime -o remember=0 Used to take about 2 minutes from the start button to having internet (Opnsense VM) running again, now it takes the best part of 10m. That section above is 3 minutes on its own. Why has adding the ZFS disk slowed the process down so much? Edit: Problem two is now I've got two NFS shares I cannot get rid of which is preventing me mounting the folders rather than the top level master drive. I think its because there are spaces? Export list for TOWER: /mnt/master/Working * <--- Not in /etc/exports /mnt/master/Mac (everyone) <--- Not in /etc/exports /mnt/user/Working-Data ....<IPs>.... /mnt/master ....<IPs>.... /mnt/cache ....<IPs>.... /mnt/disk1 ...<IPs>.... zfs get sharenfs NAME PROPERTY VALUE SOURCE master sharenfs off local master/Mac HD 2 sharenfs off inherited from master master/Mac HD 3 sharenfs off inherited from master master/Working Files sharenfs off inherited from master Edit 2: Edited: nano /etc/exports.d/zfs.exports and removed all the entries. Ran exportfs -r, and the entries are gone. But: rpc.mountd[31793]: request to export directory /mnt/master/Mac HD 2 below nearest filesystem /mnt/master ??

Took a while (disk spinning up I think) NAME PROPERTY VALUE SOURCE master type filesystem - master creation Mon Jul 24 19:08 2023 - master used 9.07T - master available 3.52T - master referenced 104K - master compressratio 1.01x - master mounted yes - master quota none default master reservation none default master recordsize 128K default master mountpoint /mnt/master local master sharenfs on local master checksum on default master compression lz4 local master atime off local master devices on default master exec on default master setuid on default master readonly off default master zoned off default master snapdir hidden default master aclmode discard default master aclinherit restricted default master createtxg 1 - master canmount on default master xattr on default master copies 1 default master version 5 - master utf8only off - master normalization none - master casesensitivity sensitive - master vscan off default master nbmand off default master sharesmb off default master refquota none default master refreservation none default master guid 5678621045851606521 - master primarycache all default master secondarycache all default master usedbysnapshots 0B - master usedbydataset 104K - master usedbychildren 9.07T - master usedbyrefreservation 0B - master logbias latency default master objsetid 54 - master dedup off default master mlslabel none default master sync disabled local master dnodesize legacy default master refcompressratio 1.00x - master written 104K - master logicalused 9.23T - master logicalreferenced 46K - master volmode default default master filesystem_limit none default master snapshot_limit none default master filesystem_count none default master snapshot_count none default master snapdev hidden default master acltype off default master context none default master fscontext none default master defcontext none default master rootcontext none default master relatime off default master redundant_metadata all default master overlay on default master encryption off default master keylocation none default master keyformat none default master pbkdf2iters 0 default master special_small_blocks 0 default Disk is now mounted and data is visible.

Cannot open master dataset does not exist. Something has occurred to me and that is the disk in 6.11.5 automatically mounted at boot (even before array start) is there a ZFS flag that I may have set and forgot that automatically mounts disks and when 6.12 imports it that flag mounts it?

zfs list: no datasets available zfs mount: blank Insightful 😄

Finally getting round to re-upgrading to 6.12 from 6.11.5 with the ZFS plugin. I've got a very basic ZFS disk: sdh 8:112 0 12.7T 0 disk ├─sdh1 8:113 0 12.7T 0 part └─sdh9 8:121 0 8M 0 part I start with it not mounted after a reboot: zpool status no pools available But I am struggling to import, followed a few guides with no success even though half way through the mount it is actually mounted and I can see the data. Sep 12 22:28:54 TOWER emhttpd: mounting /mnt/master Sep 12 22:28:54 TOWER emhttpd: shcmd (630): mkdir -p /mnt/master Sep 12 22:28:54 TOWER emhttpd: shcmd (631): mount -t xfs -o noatime,nouuid /dev/sdh1 /mnt/master Sep 12 22:28:54 TOWER root: mount: /mnt/master: wrong fs type, bad option, bad superblock on /dev/sdh1, missing codepage or helper program, or other error. Sep 12 22:28:54 TOWER kernel: XFS (sdh1): Invalid superblock magic number Sep 12 22:28:54 TOWER root: dmesg(1) may have more information after failed mount system call. Sep 12 22:28:54 TOWER emhttpd: shcmd (631): exit status: 32 Sep 12 22:28:54 TOWER emhttpd: /mnt/master: no btrfs or device /dev/sdh1 is not single Sep 12 22:28:54 TOWER emhttpd: /usr/sbin/zpool import -d /dev/sdh1 2>&1 Sep 12 22:28:54 TOWER emhttpd: pool: master Sep 12 22:28:54 TOWER emhttpd: id: 735173779397214344 Sep 12 22:28:54 TOWER emhttpd: shcmd (632): /usr/sbin/zpool import -N -o autoexpand=on -d /dev/sdh1 735173779397214344 master Sep 12 22:28:57 TOWER emhttpd: /usr/sbin/zpool status -PL master 2>&1 Sep 12 22:28:57 TOWER emhttpd: pool: master Sep 12 22:28:57 TOWER emhttpd: state: ONLINE Sep 12 22:28:57 TOWER emhttpd: config: Sep 12 22:28:57 TOWER emhttpd: NAME STATE READ WRITE CKSUM Sep 12 22:28:57 TOWER emhttpd: master ONLINE 0 0 0 Sep 12 22:28:57 TOWER emhttpd: /dev/sdh1 ONLINE 0 0 0 Sep 12 22:28:57 TOWER emhttpd: errors: No known data errors Sep 12 22:28:57 TOWER emhttpd: shcmd (633): /usr/sbin/zfs set mountpoint=/mnt/master master Sep 12 22:30:14 TOWER emhttpd: shcmd (636): /usr/sbin/zfs set atime=off master Sep 12 22:30:14 TOWER emhttpd: shcmd (637): /usr/sbin/zfs mount master Sep 12 22:30:14 TOWER root: cannot mount 'master': filesystem already mounted Sep 12 22:31:31 TOWER emhttpd: shcmd (637): exit status: 1 Sep 12 22:31:31 TOWER emhttpd: master: mount error: 1 Sep 12 22:31:31 TOWER emhttpd: shcmd (641): /usr/sbin/zpool export master Sep 12 22:32:48 TOWER emhttpd: shcmd (645): mount -t reiserfs -o noatime,user_xattr,acl /dev/sdh1 /mnt/master Sep 12 22:32:48 TOWER kernel: REISERFS warning (device sdh1): sh-2021 reiserfs_fill_super: can not find reiserfs on sdh1 Sep 12 22:32:48 TOWER root: mount: /mnt/master: wrong fs type, bad option, bad superblock on /dev/sdh1, missing codepage or helper program, or other error. Sep 12 22:32:48 TOWER root: dmesg(1) may have more information after failed mount system call. Sep 12 22:32:48 TOWER emhttpd: shcmd (645): exit status: 32 Sep 12 22:32:48 TOWER emhttpd: /mnt/master mount error: Unsupported or no file system Sep 12 22:32:48 TOWER emhttpd: shcmd (646): rmdir /mnt/master Sep 12 22:32:49 TOWER emhttpd: shcmd (647): /usr/sbin/zfs mount -a What is going on?

Is this plugin still under development? I'm still suffering from the 128MB PHP limit and there is still no way of manually triggering a disk verify. Is there an alternative plugin that can achieve the same thing and works without editing stuff for the memory limit and setting a temporary scheduled verify?

Can I do this a different way then, can I split my 2TB into 1TB for array cache (or anything that I can put VMs/Docker on) and then 1TB to be pass through’d for gaming VM instead of a PCI-E pass through? E.g. /dev/sda1 - Cache for VMs/Docker /dev/sda2 - Block device for primary gaming VM?

Not even via command line?

Great, thanks. Having read through that manual link, it does seem possible for me to create two partitions on my SSD, one 1TB in size (for "normal" cache use, docker, VMs, etc...) and then another 500GB Cache for my L2ARC (Remaining 500GB as unused for wear levelling reasons...), am I right in this? Secondly, if ZFS doesn't release the memory once pressure ramps up, I might make this 2GB initially. Then if/when I get 48/64GB of ram I might crank it up to say 8GB's worth. TBC!

6.11 before upgrade - 32-34W measured at the wall Updated to 6.12.1 - 50+W Downgraded to 6.11 - 32-34W It is like disks aren't spinning down when they say they are (I forgot to check physically if they were)... going to file a bug report.

Background: I currently have a four bay DAS connected to my Thunderbolt dock with a few different HDs in for various things, however I'm becoming increasingly frustrated that I have to be sat at my desk to do Photo Editing. Also I've been trying to do photo editing remotely (slowly admittedly) over VPN which works OK if I point Lightroom at my array (rather than "Master" disk) via NFS. So I now want to add another disk to my UnRAID server, format it ZFS, add an SSD for L2ARC and see how well it will work. Ideally want fast directory indexing and fast performance of recently added/used files. So my questions are: 1. Does (or will) UnRAID Support L2 ZFS caching, i.e. L2ARC - I think it does now via Command Line with GUI setup on its way(?) 2. I currently have 32GB of RAM, of which sometimes I'm running a Gaming VM which uses 20GB up. I'm very rarely, if ever, going to be needing fast performance whilst gaming. I assume the ARC will dynamically adjust to whatever free memory there is? 3. I currently have a 1TB SSD via SATA. I was going to replace this with a 2TB M2 (ideally SLC) drive, and make a 500GB volume for L2ARC and the rest as my normal "cache" (Docker, VMs, etc) drive, is this doable? 4. I don't have ECC memory and I don't think fast 3200 64GB ECC UDIMM kits even exist - given this mostly impacts ARC rather than data at rest is this actually a major issue. Given this disk would effectively be "Master" corruption would be annoying...! 5. If I dump 200GB of photos/videos, I assume this will be sent to the memory, then SSD first, then sent across to the HDD. This data would stay on the SSD whilst I edit it (frequently accessed) and then be pushed out when the next set of data is sent across. I.e. The latest data up to a certain size will be cached by the SSD? 6. All the data (after it's been copied from the cache) is on the HDD. So if the SSD fails for example (ignoring data that maybe be a few minutes old and hasn't been flushed to the HDD) would be there. I.e. If I connect the HDD to another machine the data would all be there. I think thats all the questions I have for now, some input on what I'm trying to achieve would be interesting! Many Thanks.

Anyone running 6.12 noticed increase in power usage? Updated two weekends or so ago and now I’m 10W higher draw all of a sudden!

So this does not, it seems, work for AMD cards because I suspect it sets BAR 2 to a size above 8MB automatically (256MB) and it seems that this causes Error 43 in guests every time. I've had to manually do it, for example for a 8GB card, 8GB/8MB: echo 13 > /sys/bus/pci/devices/0000:03:00.0/resource0_resize echo 3 > /sys/bus/pci/devices/0000:03:00.0/resource2_resize GPU-Z now reports as it being enabled, AMD reports it not being enabled but Windows 10 device manager shows "Large Memory Range". Performance... absolutely zero difference on/off, so who knows.

Some more investigation to this has found that there are indeed processes being pinned by kernal/unraid base software on the isolated cores: For example: [kworker/9:1H-kblockd] [ksoftirqd/9] Other processes using 'isolated' cores... ahci[0000:00:17.0] ahci[0000:07:00.0] xhci_hcd eth1-TxRx-x Which sits at around 5% CPU usage on this core. Based on this: https://honser.github.io/openstack/2019/11/17/Complete-Isolation-of-VM-resources/ I can add the following to syslinux to fully prevent other tasks using them: nohz_full & rcu_nocbs? Some users seem to have added this line: Anyone think of any reason why I would not add these terms? The reason I'm looking into this is due to getting stutters on the VM during gaming which appear to be processes using the VM cores briefly.

Ref this topic: For quite a while now (could very well be 2 years or more by this point as this happens with multiple Mac OS releases and Safari versions that clicking the apps tab will freeze the entire window. Also happens across three different Macs (Top of which I have now sold, so it does not seem to be related to Apple Silicon/Intel. I've tried various things but nothings worked - I simply head to Chrome to install stuff. How do we go about fixing this?