lionceau

I'm running munin on my homeserver and don't check graphs very often. I've noticed periodic bursts in power consumption at the wall and have tracked it down to munin rebuilding graphs. Not a big problem but also unnecessary because I only check graphs once a week. No problem though, I'm going to build my own docker container with this option.

could you perhaps add support for the cgi render/display strategy? that way munin only updates the site and graphs dynamically when the page is requested, not every 10 seconds via cron. http://munin-monitoring.org/wiki/MuninConfigurationMasterCGI You need the following in /etc/munin/munin.conf to generate graphs and HTML dynamically: html_strategy cgi graph_strategy cgi

Hello, Is there a way to set a share so it gets copied to the array by the mover but also stays in the cache? I have some frequently accessed important data on my cache drive and don't want the data and parity disk to spin up every time it gets read or written to. However I also don't want to leave the data only on the cache alone in case my SSD becomes defective. A mirrored cache drive is not really an option as I'm out of available SATA ports and 1TB SSDs are still expensive. I suppose I could set the shares to "Cache only" and set up a rsync cronjob that backs up the files to a second share on the Array every day but I'm wondering if there's a more elegant solution for this.

The latest news on ECC support is that Ryzen will not officially support it after all but there is a chance that it still does support it unofficially like many previous AMD CPUs do. That would require motherboard manufacturers to provide the extra traces for ECC RAM and it is unknown whether they will. A bit disappointing. I believe ECC RAM support should be standard in all devices from Servers to IoT these days. Ryzen looks like an excellent platform for unRAID provided all the small details work out.

Now that Ryzen is official I'm quite interested in it. Upto 8 Cores/16 Threads, excellent SMT implementation, ECC RAM support, IOMMU, 8 port on-CPU SATA with the X370 chipset and the high end models are half the price of comparable Intel CPUs. It seems to tick all boxes but I'm concerned whether it'll play nice with unRAID and VM PCI/GPU passthrough. Is anybody planning to test this? Update to make things a bit easier for people interested in running Unraid on Ryzen: unRAID 6.4rc7+ generally works fine with Ryzen as of July 2017. Be sure to follow your motherboard manufacturers instructions to update to the latest UEFI version with AGESA 1.0.0.6. As of right now (6.4rc6) there appears to be one bug related to CPU power-managment that causes unRAID to crash intermittently. Good news: This bug was fixed in 6.4rc7! The way to fix this is to either disable "Global C-State Control" in the motherboard's UEFI/BIOS or have a constant load on the machine (i.e. with a Windows VM idling) so that the CPU never throttles frequency down completely at idle. Either of these fixes will lead to slightly increased power consumption. The second bug is in KVM and leads to decreased video performance when passing through GPUs into VMs on Ryzen. The problem is related to Nested Page Tabling; Ryzen owners have observed decreased CPU and I/O performance inside the VM when NPT is off and decreased FPS in graphic intensive applications (i.e. video games) when NPT is on. Gaming performance inside a unRAID VM with passed through GPU can be disappointing for that reason. Some systems seem to be affected more than others. The good news is that Xen hypervisor is not affected by this problem so it looks like the bug is in the KVM software, not AMD's hardware; KVM lead developers have acknowledged the issue but are lacking time/resources and did not mention a timeframe for a fix. There are some more advanced workarounds that can improve the situation by turning off NPT and passing through entire storage controllers but don't work around it completely. If you are affected by this issue please take a look at https://bugzilla.kernel.org/show_bug.cgi?id=196409 and consider commenting to raise awareness among developers and perhaps even AMD.

Thanks for the input. I'm currently running a parity check so I can't try anything but I'll check tomorrow.

Yes thanks, I should have phrased my post differently. I've edited/replaced /boot/config/parity-checks.log with the good backup but it gets overwritten or appended with the faulty entry periodically (or whenever I reboot)

Hello, My parity check log seems to have become corrupted when restored my boot drive from a backup. It now shows one check from today and one completed from 1.1.1970: 1970 Jan 1 01:00:00|-1487685544|nan B/s|0| I do have a backup of the old parity-check.log but it seems to get overwritten when I try to replace the corrupt one with the working one. Is there a trick to make the restored backup stick?

Hi, I have a Win10 VM set up for Steam that uses a VM disk image for the steam library. The disk image is rather large and thus has to be stored directly on the Array. My problem was that Windows 10 would continuously issue small reads/writes while idle. This spun up both the parity and the array disk for only 1 or 2 reads/writes as per unraid dashboard. Since the drive only "sees" the large 1TB disk image, small accesses don't get cached so my target was to eliminate every single one of them. I think I've fixed the problem now with these steps. I'll update this list as I find more services/programs accessing the drive. Please be aware that disabling services and editing the registry can break functionality so be sure to keep a backup and don't follow this unless you know what you are doing. - disable Superfetch and Prefetch - in "HKEY_LOCAL_MACHINE\Software\Microsoft\Windows\CurrentVersion\Policies\Explorer" create a new DWORD value named "NoLowDiskSpaceChecks" and set to value "1". This disables regular checks by explorer.exe for low disk space. - set Steam to autoupdate in a limited timeframe, again preferably at night just before the mover spins up disks anyway. - disable the following services: ncbservice (network connection broker), pcaservice (program compatibility assistant service) for VMs with Nvidia cards only: - disable NVtmrep in the task scheduler, it's a Nvidia driver component that seems to scan all disks once a day for telemetry/debug purposes. - reschedule NvProfileUpdaterDaily to run at night just before the mover After these steps my disks stay spun down even though Win10 and Steam are running idle and the VM disk image is mounted. If you still have problems I can recommend Windows Process Monitor and Process Explorer for debugging the random read/writes. update: I've given up on this, there always seems to be some service, program or task that randomly wakes up the array for no reason. The way I've solved it is with two diskpart.exe batch scripts, one to put the virtual disk offline and one to put it back online inside Win10. online.bat runs on startup and via a manual shortcut inside steam offline.bat runs after the computer is idle for 15 minutes this way i can manually spin up the disk via steam remote streaming and can be sure that the disk is offline the rest of the time.

Hello, Many systems require a GPU dump to be specified so VM GPU passthrough works without Error 43 as described in this video: http://lime-technology.com/forum/index.php?topic=52960.0 It would be nice to have an (optional) text box/file picker box for this in the advanced VM PCI passthrough settings GUI so we don't have to copy-paste this into the XML whenever we change a VM setting. The XML tag is <rom file='/mnt/user/path/file.dump'/>

EVGA 1060 SC 6GB (non-SC version posted in the original topic gave me Error 43) P/N 06G-P4-6163-KR BIOS 86.06.39.00.62 Works perfect now, great performance and runs very cool as I was able to undervolt it to 0.903V at only 30 Mhz below stock clockspeed. Never goes above 65% power limit. http://gpuz.techpowerup.com/17/02/02/ega.png https://www.expirebox.com/download/707fd4d24328be8bc174bc6136d9fcd5.html

Hello, I've adjusted unRAID to my liking and am at the point where I won't change much in the future. My cache drive is unmirrored because of SATA port restrictions and my preference for XFS. My docker containers, VMs, VM isos, and appdata are only on the cache drive. Is it possible to stop the array and execute dd if=/dev/sdb of=/path/to/external/disk/image.img to save an image of the cache drive to an external drive so I can do the reverse in case the cache drive dies and have a "known good" status without having to reconfigure everything? CA backup already saves appdata to the array every month but that doesn't include VMs and the other stuff. I feel like being able to just restore a raw diskimage using dd on my laptop (with an USB 3.0 dock) would be the fastest and easiest recovery method in case of a failure.

I'm aware of that but I'm not sure how that relates to my question. By "UI" i meant the VNC Crashplan UI, not Unraid. That being said crashplan's network architecture doesn't seem to play nice with port redirection at all.

Has anybody managed to make this work with a non-default listening port? Whenever I try to change the default 4242 in the UI it won't accept the port and instead applies a seemingly port within 10 ports of the one I picked. For example if I enter "Listening Port" 5000 and press apply, crashplan accepts Port 5006 instead. I've tried this with many completely random ports, there's no way _all_ ports on my machine are taken. These random port won't even stick through reboots. Setting the service port in the unRAID docker ui also does nothing. I noticed that the name of the variable changes when the port in the docker UI gets changed. e.g. the container normally starts with "-e TCP_PORT_4242=4242" but when you change it to 5000 it starts with "-e TCP_PORT_5000=5000" Is that intended? by the way, do you guys run host or bridge networking?

This won't work with unRAID. The required kernel modules aren't present. Also isn't this for a laptop/desktop system running on a battery? There are some power managment related tweaks that are useful for normal desktops/servers. The powertop tweaks (more aggressive pci power savings, etc) lower my idle wattage from 21 to 15 watts with open media vault. Its not the end of the world but it would have been nice to have. Unraid idles at 31 watts, presumably because I can't figure out how to enable the kernel mode setting for the intel GPU and powertop is not available.

can you please add powertop to the nerdpack? that would be very useful to reduce idle power consumption.

Hello, my Dell T20 with a Xeon 1225v3 uses 30W idle with all drives spun down. The same machine idles at 15W with another debian based NAS distro. Apparently it is vital to enable Kernel Mode Setting so that the iGPU can enter sleep states and save power. Does anybody know if unRAID uses KMS by default or how I can activate it?

Hello! I'm trying to make myself a docker image which I can use to access a certain share remotely via SSH/SFTP without having to access it through unRAID's sshd. Unfortunately I'm having a very hard time grasping Docker and figuring out the easiest "workflow" to get the container set up perfectly from a dockerfile. Is it viable for me to create one persistent docker container, configure it to my liking, save it as an image and just update that with apt-get as I would with a VM? From what I gather this is frowned upon by the community but I don't plan to move or distribute the docker image. Normally I'd just use a VM for this but I found that the io performance of 9p virtio is much worse than what a docker container can do.

Yes thank you, I checked that before making the VM. Unfortunately they are all socket based and none of them include ssh for tunneling. I'd like to use unison over ssh (exposed to the internet) so I can sync remotely without having to resort to a VPN. Syncing with the socket method over the internet in plain text is a bad idea.

Hello, I'm trying to run unison (a sync tool similar to rsync) on unRAID. I'm not talented enough to package it into a docker and keep it maintained. Instead I set up an Ubuntu VM, configured everything up to my liking and mounted the target share via virtio 9p. Everything worked out of the box but performance is far lower than what I hoped. My dd zero throughput from the Ubuntu VM to the SSD cached share is ~70 MB/s. The same from the unRAID host returns ~430 MB/s. A (io-heavy) sync which took 0.960 seconds on my C2D Mac Mini + 8 year old Drobo now takes 22.638s seconds on the new box and utilizes 60% of all CPU cores for the duration. This is worrying as I've got a cronjob that executes this sync every 15 minutes on all of my machines. CPU usage inside the VM is actually low. Dell T20 + UnRAID 6.2.4 Trial e3-1225v3 16GB ECC 512GB 850 Evo Cache drive 3x8TB WD Reds My fstab entry is as follows: unison /unison 9p trans=virtio,version=9p2000.L,nobootwait,rw,_netdev Is Virtio 9p just the wrong tool for this job? Is there a faster alternative?

How does this container relate to the official Plex Docker container going forwards? Starting from a fresh unRAID install, should I pick this or the official Plex image?

I'd like to request unison + ssh. http://www.tecmint.com/file-synchronization-in-linux-using-unison/ It is a very stable crossplatform file syncing utility with excellent conflict management. I use it as a replacement for BTsync. It tunnels through SSH so it's quite safe. Since it's feature finished and mature it doesn't get updated often, so the maintenance requirements past the initial development should be minimal.

I'm no expert on the matter but surely it can't be a good idea to use SMR drives as parity? They have two levels of cache (128MB DRAM + 20GB on-disk) managed by a firmware that's optimized for archival (few, large sequential writes). The screenshots suggest Parity workload is the exact opposite.

This is amazing, I hope it gets updated with unRAID as new versions come out. The unattractive unRAID webinterface was one of the main things why I refused to even give it a try in the past. Once I tested it, I quickly understood why unRAID is so highly regarded and quickly saw past the interface design but this CSS shows that it could be perfect.

That's a pity. The drives are just out of warranty but I don't think WD would do anything about the high LC count. I've ordered 2x8TB WD Reds. One as parity, one as storage. I'll keep the two "good" WD Reds in the new array and use the older ones as a monthly backup.