Jump to content

rragu

Members

  • Joined

View Profile Find content

Posts posted by rragu

    I followed the IBRACORP guide to setting up Tailscale. I then tried accessing the unRAID web UI from my iPhone to test after turning off WiFi.

     

    When on LTE and connected to Tailscale, I can successfully ping 100.x.y.z. When I try to navigate to 100.x.y.z in Safari, I get redirected to "xxxxx.unraid.net". But, the request times out before I can actually access the login page.

     

    I deleted the My Servers plugin from unRAID and turned off SSL for the web UI in Settings/Management Access. And then I was able to access the unRAID dashboard on my phone via Tailscale while on LTE.

     

    My questions are:
    1) Is turning off SSL for the web UI required to be able to access the dashboard via Tailscale? If not, I'd be interested to hear if you had to do anything special to get the web UI SSL and Tailscale to play nicely together.
    2) Assuming I only access the web UI remotely via Tailscale (i.e. NOT via exposing directly to internet or reverse proxy), is using SSL on the web UI even necessary?
    3) Would I be missing out on something by not using SSL (other than a nice lock icon in the browser)? Is there any appreciable security risk (either in theory or in practice)?

     

    Thanks!

    • Like 1

    ·

    Edited by rragu

    EDIT: SOLVED

    Looks like my vBIOS was the issue! Despite following SpaceInvaderOne's video/script to dump the vBIOS from my card, that vBIOS doesn't appear to work. it was only after I used one of the compatible vBIOSes from TechPowerUp that I was able to see the following in the VM's Device Manager:

    Device Manager.png

     

    I also undid the ACS override and stuck with i440fx as Machine Type

    As with DemoRic above, I get the following error: "Fatal error: Cannot redeclare _() (previously declared in /usr/local/emhttp/plugins/parity.check.tuning/Legacy.php:6) in /usr/local/emhttp/plugins/dynamix/include/Translations.php on line 19" although I get it once the Array is started and I'm logged into the dashboard (not only when the array is stopping). I don't think I got it before installing v2021.09.10.

    Posted ·

    Edited by rragu
    clarity

    I've been trying and failing to get my graphics card passed through to a Windows 10 VM for a few hours now no matter what I try. I'm going to need some help to go any further.

     

    I've been largely following this guide on setting up remote gaming

     

    Details/Settings:

    GTX 1080 Ti is in the motherboard's top slot (so I guess that makes it the primary GPU?)

    ACS override: set to Both

    VFIO: Both graphics and sound devices stubbed via Tools>System Devices

    Boot: Legacy boot

    W10GamingVMDetails.png

    W10GamingVMXMLDetails.png

     

    After adding the graphics card with settings as detailed above and booting the VM, Device Manager doesn't recognize any nVidia card as being installed. All I see under Display Driver in Device Manager is Microsoft Basic Display Adapter and Microsoft Remote Display Adapter.

     

    If I go ahead install the nVidia drivers anyway, I then get Code 43 (which I suppose isn't surprising at all if Windows doesn't recognize the GPU in the first place).

     

    Ideas on how to move forward from here would be very much appreciated. Thanks!

    For anyone stumbling across this post having the same issue, this problem *appears* to have been solved by simply updating my BIOS.

     

    From some cursory Googling, the problem appeared to be related to overclocking my memory (to be clear, I was only running it at the XMP-rated 3600MHz speed). I had already run a memTest on the memory sticks which brought up no errors.

     

    So, I simply updated my BIOS to see if that would solve the problem. I haven't had any Machine Check Events warnings since.

    • Like 1

    First off, thank you @Sycotix for your Authelia CA container as well as your video series on YouTube. Very helpful and detailed!

     

    I've set up Authelia using a combination of your video and this blog post by Linuxserver. I mostly followed your video except for the end where I used SWAG instead of NPM.

     

    I've tested Authelia by protecting two endpoints: Syncthing and Tautulli.

     

    A few questions:

    1) When I go to https://syncthing.mydomain.com, I get a distorted Authelia login page (please see attached images), whereas when I go to https://tautulli.mydomain.com, I get the usual Authelia login page. This is the case on desktop Firefox, Chrome, and Edge. I don't suppose you've seen this before? Any ideas as to why this might be? The distorted page is still functional (just not as pretty).

    EDIT: tried on mobile Chrome (iOS) and mobile Safari. For both mobile browsers, both Syncthing and Tautulli give me the distorted Authelia page.

     

    2) In any case, once I login, I get to another login prompt. Obviously this is from the authentication I enabled before Authelia was set up. So, now that Authelia is protecting these services, am I good to just disable the "internal" (for lack of a better word) authentication for these services?

     

    2a) I disabled the basic GUI auth for Syncthing. And while Authelia of course still protects Syncthing, I do now get a bright red warning message from Syncthing that I need to set GUI authentication. Is there any way to make Syncthing aware of Authelia or link them in some way so that the warning message goes away?

     

    3) For the majority of my reverse-proxied services, I will probably be the only one who needs to access them. But for certain services (e.g. Ombi) where I would have multiple users, how do I set it up such that userX and userY logging in via Authelia automatically signs in userX and userY, respectively, to the desired service?

     

    Thanks for any and all help!

    Authelia-distorted.png

    Authelia-regular.png

    Posted ·

    Edited by rragu

    So, I'm in the process of setting up my unRAID server when I got a notification regarding Machine Check Events.

     

    I've attached the diagnostics. The relevant part of the syslog appears to be:

    Quote

    Jun  4 22:53:27 TheRamsServer kernel: mce: [Hardware Error]: Machine check events logged
    Jun  4 22:53:27 TheRamsServer kernel: [Hardware Error]: Corrected error, no action required.
    Jun  4 22:53:27 TheRamsServer kernel: [Hardware Error]: CPU:0 (17:71:0) MC27_STATUS[-|CE|MiscV|-|-|-|SyndV|-|-|-]: 0x982000000002080b
    Jun  4 22:53:27 TheRamsServer kernel: [Hardware Error]: IPID: 0x0001002e00000500, Syndrome: 0x000000005a020005
    Jun  4 22:53:27 TheRamsServer kernel: [Hardware Error]: Power, Interrupts, etc. Ext. Error Code: 2, Link Error.
    Jun  4 22:53:27 TheRamsServer kernel: [Hardware Error]: cache level: L3/GEN, mem/io: IO, mem-tx: GEN, part-proc: SRC (no timeout)

     

    Can anyone please help me to understand this output? My server's component details are in my signature.

     

    Thanks!

    server-diagnostics-20210604-2332.zip

    8 hours ago, ich777 said:

    But keep in mind that you don't have to extract the BIOS from the card anymore. :)

    Sorry, just to make sure I'm understanding you right:

    I don't need to do anything to the 1080 Ti primary GPU other than bind it to VFIO via System Devices? I don't need to specify the vBIOS in the W10 VM's config/XML etc.?

    5 minutes ago, ich777 said:

    Go to Tools -> System Devices -> Click the checkboxes for the 1080Ti (Graphics and Audio) and click on Bind to VFIO on Boot, this should make the card invisible for Unraid and the plugin so that ot ia reserved for exclusive access in the VM.

     

    Thanks!

     

    Are there any possible issues that could occur as a result of stubbing the primary GPU (just wondering if there is something to look out for)?

    Hi, not entirely sure if this is the right place to post this but here goes:

     

    My setup:
        - CPU: R9 3900X
        - Motherboard: Asus Crosshair VIII Hero
        - PCIe x16 Top Slot: GTX 1080 Ti
        - PCIe x16 Second Slot: Quadro P2000
        - PCIe x16 Third Slot: LSI 9207-8i
        - Running unRAID 6.9.2

     

    What I want to accomplish:
        - Pass through the primary GPU (1080Ti) for a W10 VM for gaming
        - Use the secondary GPU (P2000) for Plex/Emby hardware transcoding

     

    From what I understand, I need to:
        1) Dump the vBIOS (following SpaceInvaderOne's video) for the 1080Ti since it's an nVidia GPU in the primary slot
        2) Install this nVidia plugin to use the P2000 for hardware transcoding in Docker

     

    My question is:

    Apart from 1 & 2 above, is there anything special I need to do to accomplish my goals (e.g. stubbing the primary GPU or something like that)?

     

    N.B.: if switching the GPUs (i.e. put the P2000 in the primary slot) would somehow make things easier, unfortunately I can't. My 1080Ti is a 2.5 slot card and there isn't enough clearance between the second PCIe slot and the LSI HBA in the third PCIe slot.

    On 8/29/2020 at 5:25 AM, mgutt said:

    1.) Check the destination documentation of rclone. Every destination has its own special parameters and behaviour like Google Drive has its own, too:

    https://rclone.org/drive/

     

    2.) Do not copy/sync to the mount of the destination like /mnt/disks/gdrive. Use only the rclone alias like gdrive:subfolder. rclone does not work properly if the target is a local path. Example: If you use the mount path and your destination is a webdav server it does not presere the file modification time.

     

    3.) If it does not preserve the file modifcation time it must use checksums to verify that the destination file is the same (or not) before being able to skip (or overwrite it). This means it downloads and calculates the checksums (high cpu usage) and uploads at the same time (rclone uses 8 parallel running --checkers for this).

     

    4.) There are minor performance differences between sync and copy as they act a little bit different, but finally this should not influence the transfer speed of files:

    sync = deletes already existing files from destination that are not present on source

    copy = ignores existing files on destination

     

    Conclusion:

    Check if your transfer preserves the timestamps. If yes, then disable the checksum check through --ignore-checksum (if you are ok with that) and test different --checkers and --transfers (4 is the default) values until you reach the best performance. For example I use "--checkers 2 --transfers 1" for my Nextcloud destination as there was no benefit uploading multiple files and finally it does not raise the total transfer speed if I have two parallel uploads instead of one.

     

    My command as an example: 

    
rclone sync /mnt/user/Software owncube:software --create-empty-src-dirs --ignore-checksum --bwlimit 3M --checkers 2 --transfers 1 -vv --stats 10s

     

    Thanks! I lowered the checkers to 2 and transfers to 1. Combined with a chunk-size of 256M, I get the same ~80MBps with half the CPU utilization as before, even without --ignore checksum

    2 hours ago, Stupifier said:

    The biggest issue with writing to the rclone mount is just flat out reliability. People pretty much ALWAYS complain about it. Either it gives errors, or its slow, or it doesn't copy everything you told it. Pretty much, it isn't something you'd wanna use/trust. I know its convenient......sorry.

    It is such a well-known thing that there are a ton of very popular scripts around GitHub that basically monitor directories and perform regular rclone sync/copy jobs for you in the background on a regular schedule (like every 2 minutes or whatever you set). One such script is called "Cloudplow". Very well documented/mature. Easy to find on github.

     

    Now, rclone mount is absolutely remarkable as a read source. It is excellent for reading.

    Just tried out "rclone copy"....the difference is night and day

     

    Test files: 4 files (total of 12.3 GB; between 2.3-3.6GB each)

    Average transfer speed using rclone mount: 19.4MB/s

    Average transfer speed using "rclone copy": 60.9MB/s

    Average transfer speed using "rclone copy" and chunk-size 256M: 78.1MB/s

     

    The only drawback is heightened CPU/RAM usage but I'm sure I can manage that with a script like you mentioned.

     

    Thanks very much for all your help!

    10 minutes ago, Stupifier said:

    https://rclone.org/commands/rclone_mount/

    https://rclone.org/flags/
    CTRL+F "chunk"

    And FWIW, It is not the best idea to Write to an rclone Google Drive mount. Its just a widely known tip....those rclone mounts are more geared towards reads. The If you want to write something to a Google Drive remote, doing it manually using "rclone copy" command and the flags link I provided has a flag to designate chunk size.

    Also, even though you have Gig FIOS connection.....you still may not saturate your connection using rclone/google. That might just be the speed you get with Google. It's different for everyone.

    Thanks! I'll look into the resources you posted.

     

    As for not writing to the rclone Google Drive mount, (1) it's a slightly more widely known tip now 😅, (2) while I'll switch to using "rclone copy", is there any particular negative effect to transferring data to Google Drive in the way I've been doing (e.g. data loss/corruption) or is it just lower performance?

    Hi, I recently set up rclone with Google Drive as a backup destination using SpaceInvaderOne's guide. While archiving some files, I noticed that my files were being uploaded at around 20MBps despite having a gigabit FiOS connection. Based on some Googling, I'm thinking increasing my chunk size might improve speeds.

     

    But how do I go about increasing the chunk size? I've attached my rclone mount script if that's of any help.

    Also, how does this affect the items I have already uploaded (if it affects them at all)?

    3 hours ago, jonathanm said:

    To expand on what @Energen said, unless you are trying to distribute the weight so that it's less concentrated in the case and instead making many lighter packages, then removing the hard drives SHOULD be unnecessary. That assumes they are mounted properly, obviously if they are only held in with one screw on a makeshift piece of scrap, then remove them. Bare hard drives are especially vulnerable to impact, the simple act of setting one bare drive on top of another can exceed the design specs for instantaneous G loading. If it made a clack noise, you probably went over the limit.

     

    However... your heatsink should be either removed or secured. I've seen multiple instances where a heatsink came off the mounts and played pinball inside a tower. In one shipment from Alaska to the southern US, the end result was not pretty, I think the only thing salvageable was the DVD drive.

    Well it'll be a seven hour drive. Personally, I'm willing to completely waste an hour of my time to gain that bit of peace of mind (even if it might be illusory 🤷‍♂️). Besides, what with quarantining, each hour of my time is suddenly much less valuable...

     

    As for the heatsink, I use an AIO (probably also overkill for this use-case; but I had it left over from another build). I'm thinking that an AIO shouldn't need to be removed, as it's not a hunk of metal like a NH-D15 etc.?

    Posted

    I'm planning to move my server from my parents' house to mine.

     

    So far I'm planning on:

    - running a backup via Duplicacy and the Backup/Restore Appdata plugin (I already do this daily and weekly respectively)

    - running a parity check before the move

    - noting which HDD is connected to which SATA port

    - removing the HDDs and expansion cards and packing them safely for the drive

    - reinstalling the components post-move in the same manner they were pre-move

    - running another parity check to ensure there was no damage to the HDDs as a result of the drive

     

    A few questions:

    1) Is there anything else I should be considering?

    2) Currently, my server has a DHCP reservation of 192.168.x.y; the DHCP reservations at my house follow a slightly different scheme. Apart from simply creating a new reservation for the server on my router, is there anywhere within unRAID I need to manually update?

    3) I run a number of reverse-proxied services on unRAID. Since I run cloudflare-ddns, I take it Cloudflare will automatically be updated with the new public IP (i.e. I don't need to do anything or reinstall LetsEncrypt etc.)?

     

    Thanks for any help/advice!

    On 4/15/2020 at 6:00 AM, bonienl said:

    The way it is implemented in 6.8 and later, is that the passphrase is no longer stored in a file, but internally cached by the emhttpd process.

     

    This makes the passphrase invisible/non-retrievable once entered at startup of the array.

     

    When using a keyfile, this file is copied to /root prior to starting the array. The option exists to delete the file from /root after the array is started/

     

    My standard disclaimer: I only know enough to break things that I don't know how to fix...

     

    I've written my go file such that at boot, I get my array passphrase via AWS Secrets Manager and write it to /root/keyfile. unRAID then uses /root/keyfile to unlock/startup my array. I've been manually deleting my keyfile after startup.

     

    The aws-cli command I use for the procedure above retrieves a string, not a file. So, is it possible to use the output of this command as the passphrase rather than writing it to a keyfile first?

     

    Thanks!

    ·

    Edited by rragu
    Problem solved

    38 minutes ago, atribe said:

    Probably the cost of switching the repo. See the bottom left of you screenshot? Libray/telegraf. You customized it. You might be able to go into the file system where that template info is stored and put the icon there.

    Thanks!

     

    EDIT: Solved both issues as described in my post above.

    ·

    Edited by rragu
    Problem solved

    I recently set up Telegraf+InfluxDB+Grafana (+HDDTemp).

    In order to get Nvidia GPU stats, I changed my telegraf repository from alpine to latest. Life was good.

     

    However, I've just noticed that, in the Docker tab of unRAID, the Telegraf icon is missing and the docker container name is no longer a link to edit the template. If I click the docker icon, I only have the options of Console, Start, Stop, Pause, Restart, and Remove. Telegraf-DockerView.thumb.png.4f0d932f5925862ec3532fc9ca026405.png

    My Grafana dashboard is still populating properly. So, Telegraf still appears to be doing its job.

     

    That said, can anyone help me figure out what I broke? Or is this just the price of switching the repository (I would switch back to confirm this but as stated, I can't edit the template)?

     

    EDIT: Found the problem...somehow the my-telegraf.xml file at /boot/config/plugins/dockerMan/templates-user was deleted. Not a clue how that happened as I don't make a point of rooting around in /boot unnecessarily. In any case, thanks to the CA Backup/Restore Appdata plugin, I copied the file back to /boot. A refresh of the Docker tab shows I'm back to normal (icon and all).

    I installed LSIO's code-server container and set up a reverse proxy using the LSIO LetsEncrypt container.

     

    My issue is: on Firefox, once I go to code.mydomain.com and log in with the password I set in the code-server container, I just get a blank page. However, the same site works perfectly fine on Chrome and Edge.

     

    I know I'm being kinda vague. But any ideas as to why this might be or what I should start checking for?

    ·

    Edited by rragu
    clarity

    10 hours ago, GuildDarts said:

    Thanks for letting me know ^_^

    Just pushed an update. I was unable the get the error myself (on beta), but did fix what was causing the error (thanks for the logs :))
    hope this fixes it for you

    Don't know if this is expected behavior: I just updated Docker Folder to v2020.05.03 (not that I was having any issues with the previous version); I then switched to the Docker tab to see all my folders were deleted. I'm now back to the original list of containers.

     

    Edited to add: While I was able to update containers, I cannot stop them (i.e. I clicked the container icon, clicked Stop and nothing happened). Uninstalled Docker Folder and I am able to start/stop containers again.

    Hi,

     

    I installed recordings-converter and tested it out on a couple of files. Conversion logs are attached.

     

    A few questions:

    1) I think I've set it up correctly to use my GPU. Is the attached configuration correct?
    2) While the docker was doing its work, I did see load on the GPU and an ffmpeg process. But there was still 5-10% load on the CPU (and no other docker should have been using that much CPU resources at the time). Is this expected behavior?
    3) Does the container work on multiple files sequentially or concurrently? So far I've tested the container only on folders with one recording in them. When I switch to my entire recordings folder, I just want to make sure my other containers/VMs etc aren't affected. 
    4) I am using my GPU for hardware transcoding in Plex and Emby as well. It is my understanding that as long as my GPU isn't passed through to a VM, multiple Dockers can use the GPU concurrently. Is my understanding correct?

     

    Thanks for your help and a bigger thanks for all your work on these containers. Definitely makes my life easier!

    Recordings-Converter Configuration.jpg

    postProcess.18-03-2020-0712.log postProcess.18-03-2020-0839.log