aim60

I ran across an article whose author claimed that on recent versions of qemu, discard=unmap was functional on virtio disks. So I did some testing. I have not run all permutations, but have tested: Unraid 6.9.2 UEFI and Seabios VMs Q35-5.1 and i440fx-5.1 VMs Windows 10 and Windows 11 with the latest virtio drivers, virtio-win-0.1.196.iso Ubuntu 21.4 raw and qcow2 vdisks vdisks on XFS and BTRFS disks vdisks with and without the copy-on-write attribute set In all cases, the virtio disk was functionally equivalent to a virtio-scsi disk. On file deletion, windows unmapped blocks immediately. Linux unmapped blocks after running fstrim. https://chrisirwin.ca/posts/discard-with-kvm-2020

Changed Status to Solved

Missed it. Thanks Changed Status to Solved

@JorgeB, Isn't it possible to remove the drive from the pool without reconnecting it first (like if it was completely dead)? That way, I wouldn't have to worry about the small possibility of corruption to the few files that aren't cow.

I am getting write_io_errs, read_io_errs, and flush_io_errs on one of the devices in my cache pool. The pool is raid1, with enough unused space that I can maintain redundancy without the problem drive. I would like to remove the drive with errors. Please confirm that this is the correct procedure. Thanks tower7-diagnostics-20210217-1146-B.zip BTRFS Info.txt

Been using a CP1500PFCLCD for years with the unraid apc daemon. No problems, except you should set “Turn off UPS after shutdown” to no. It doesn’t work correctly with the apc software. Tried with NUT several years ago and it didn’t work work with it either.

Adding a container variable of type path, with container path and host path of /usr/bin (and access mode of read only for security) , would substitute the docker's /usr/bin with unraid's /usr/bin. However, this would probably have negative consequences. The files are probably be from different linux distros. I tried it with a random docker, and it wouldn't even start.

Install the netdata docker and take a look at the template variables, including what you see when you press the Edit buttons. Some of the system variables are defined read-only.

For those of us with removable drive bays, should we be doing anything after unmounting a drive, before physically removing it?

I'm running 6.8.1 with a 3-drive raid1 cache pool. I've been getting Errors on Cache Pool messages, and many of the following in the logs: BTRFS error (device sdf1): error writing primary super block to device 1 kernel: sd 7:0:0:0: [sdf] tag#15 UNKNOWN(0x2003) Result: hostbyte=0x04 driverbyte=0x00 kernel: sd 7:0:0:0: [sdf] tag#15 CDB: opcode=0x2a 2a 00 1d 9b af c0 00 00 58 00 The device still shows a green ball on the Main tab, but smartctl returns Short INQUIRY response, skip product id A mandatory SMART command failed: exiting. To continue, add one or more '-T permissive' options. I'm assuming that the drive needs replacement. My only use of cache is VM's and Dockers, and I have disabled both services, as well as temporarily disabling the mover job. I would like to remove the cache pool from the configuration until I can obtain a replacement. I assume that the procedure is to stop the array and unassign all of the cache drives, and start the array. Then when the new drive arrives, reassign the 2 good drives and the replacement, and start the array. Please confirm my procedure. I only have remote access to the server for the next few days. Thanks tower7-diagnostics-20200208-1324B.zip

I just purchased two WD80's. To my surprise, what arrived were WDC_WD80EFAX-68LHPN0's. I thought all new stock was air filled. Helium is a plus for me, they run a few degrees cooler.

Its for this kind of issue that binhex isolated preclear in a docker.

Once the plexpass is associated with your plex account, connect to your plex server using the web interface. Logout of your plex account, then login again. You’re done. You now have all the features of plexpass. Binhex’s plexpass docker gives you access to beta releases, not additional features.

Was waiting for a 6.8.1 stable that fixed the problem Settings-->VM Manager-->DIsable VM function does not work without stopping the array. If an 6.8.1-RC seems stable, and it is the only way to fix the problem, I'll probably jump early.

I'm running 6.6.7. After running preclear, I am attempting to format a new 8TB drive. I updated UD, installed UD Plus, and enabled Destructive Mode. Clicked the red X, and entered Yes to remove the partition. The log says that the partition is being removed, but it's not. The red X comes back. fdisk -l shows that the partition still exists. Running parted returns "parted: error while loading shared libraries: libreadline.so.8: cannot open shared object file: No such file or directory". I tried removing UD and UD Plus, removing their plugin folders in flash/config/plugins, rebooting, and reinstalling. Same result. Any ideas? tower7-diagnostics-20200107-0943-M.zip

They changed their development plans. AES-NI is no longer a future requirement. Mentioned here: https://www.netgate.com/blog/pfsense-2-5-0-development-snapshots-now-available.html

Another option is the Virtual Machine Wake on Lan plugin, and a wol app on your phone or tablet. I’m using the “WOL” app on an ipad.

For me, building with removable drive bays, then keeping my hands out of the box unless absolutely necessary, has added more to the reliability of my hardware than any other factor. +1 for using server class hardware.

Unless you really need it, make sure UPnP is turned off on the router.

I'm recording to a local drive on the MC pc. Weekly, I am moving the recordings to unraid. They never exist in both places at the same time. After moving recordings, I need to restart the Windows Media Center Receiver Service so MC registers the change in the location of the recordings. Keep in mind that with your method, you are asking unraid to do something that it was never designed to do - receive data on a real-time basis. The more simultaneous recordings that you do, and other things you may have running on on the server (plugins, dockers, VMs), the more difficult it will be for unraid to keep up. The forums on www.thegreenbutton.tv have several threads on user experiences recording to a NAS. In the future, I am going to attempt to run Win7 MC as a VM on unraid. And I'll either direct recordings to a dedicated Unassigned Devices drive, or pass through a physical drive to the VM for recordings. In my house, all sh*t breaks loose if recordings are corrupted.

I've always been reluctant to try your method, except on my test MC server. Especially with reports of array slowness in 6.7.x. I record locally, then weekly, copy the recordings to Unraid, and remove them from local storage. At the very least, consider limiting your recordings share to one disk, and have the link reference \\server\diskx\sharename rather than \\server\sharename, bypassing the overhead of the Fuse file system. There's no problem having MC use unc paths to previously recorded content. Its only the recording drive that needs the mklink trick. Nothing else comes to mind about using MC with Unraid.

The following is from Windows 7 MC. In MC, go to Tasks | Settings | TV | Recorder | Media Libraries | Remove folders from library. Verify that you do not have a reference to the recordings share on Unraid. If you do, delete it. Media Center should only know about your local drive, which is linked to the share. It won't show up in this list because its the recording drive). If MC knows about both, you'll show the duplicate recordings.

I use the Recycle Bin plugin for accidental deletions. Works well, but only catches deletions via SMB, not through the console or a file manager like krusader.

@binhex, consider adding this information to the first post. It will help new users decide which docker to use.