Howboys
-
Posts
105 -
Joined
-
Last visited
Content Type
Profiles
Forums
Downloads
Store
Gallery
Bug Reports
Documentation
Landing
Posts posted by Howboys
-
-
Here's the latest one that I've been unable to modify (this time due to shfs?):
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME shfs 12255 root 188r REG 9,4 4069601929 6442498010 file.mkv
-
Anyone got ideas? Still running into it every once in a while.
-
No windows. My other system that copied the file to the share is Ubuntu.
Basically, 1 outlet of 20 or so files copied to the share end up like this.
-
I've had this issue for a while now. Basically, I have a single share mounted on my server with smb. After a large enough copy (and sporadically, not always), if I try to mutate the file (move, delete etc), I get "Device or resource busy". Today, I was able to find out what on the unraid box is keeping the file busy:
lsof $FILENAME COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME smbd 25896 howbowz 443w REG 0,45 7039913529 648799827760513201 $FILENAME
To be clear, I ran the above in unraid's terminal.
So now my questions:
- What's going on here with smbd keeping the file open for a long, long time?
- How do I prevent it?
-
In that case, what's the ideal setup for me if I don't want to copy (use hard-links)? Even if I were to have qbit downloads in its own share, that wouldn't really solve my problem would it?
-
Would really like to understand how this works, and if I should re-do my shares.
-
This is my directory structure:
* Movies:
* `Media/Movies/Movie_Name/movie_file.mp4`
* `Media/Movies/Movie_Name/movie_file.nfo`
* `Media/Movies/Movie_Name/movie_file.srt`
* `Media/Movies/Different_Movie_Name/movie_file.mp4`
* `Media/Movies/Different_Movie_Name/movie_file.srt`
* qbittorrent* `Media/qbittorrent/incomplete/Show0/Season X/episode1.mp4`
* `Media/qbittorrent/incomplete/Movie2/movie_file.mp4`* `Media/qbittorrent/Show1/Season X/episode1.mp4`
* `Media/qbittorrent/Movie3/movie_file.mp4`With a split level of 2, `Movie_Name` and `Different_Movie_Name` directories may be on different disks but each movies entire contents will be on the same disk.
What happens is that Sonarr/Radarr send a torrent to qbit to download to `Media/qbittorrent`. Qbit downloads to `Media/qbittorrent/incomplete` and then moves the completed download to `Media/qbittorrent`.
Now, with radarr and sonarr, they "copy" files from download client to my actual share using hard-links (not actual copy). How would that split work with this hard link? Will the split at the time of directory/file creation in qbittorrent be the permanent one? In other words, if the initial download happens in `Media/qbittorrent/incomplete`, does that means that the entire `Media/qbittorrent/incomplete` will be on the same disk, and any hardlinks created to that will not actually move the file to another (more appropriate disk)?
-
That makes sense. Thank you - it was my first time replacing a disk - hopefully iy finishes without any issues.
-
I just replaced a disk with a larger one and I have the triangle as well with "DEVICE CONTENTS EMULATED" and the array is "Parity-Sync/Data-Rebuild" right now. Is the emulation and the yellow triangle expected until the sync fails?
-
Oh man On Deck would be great. Thanks so much work making this script.
-
6 hours ago, johnnie.black said:
Avoid Marvell controllers, if two ports are enough get an Asmedia/JMB582 two port controller, if you need 4/5 get a five port JMB585 controller.
Thanks for the recommendation. It would be nice to get 4 ports at least to future proof. But those look pretty pricey. Do you have specific products in mind I can look at?
-
I shouldn't clarified - I'll move my existing HDDs to the new SATA expansion card, and hook up SSD to the Mobo
-
So I have a X8SIL-F mobo and have maxed all 6 existing SATA ports. I want to add another data drive, and start using a SSD for a cache.
The board has 2 (x8) PCI-Express 2.0, 1 (x4) PCI-Express (using x8 slot). So I'm thinking of getting a pcie expansion card to add more SATA ports. Would something like https://www.amazon.com/gp/product/B07SZDK6CZ/ or https://www.amazon.com/IO-Crest-Controller-Non-Raid-SI-PEX40064/dp/B00AZ9T3OU work? Is there a cheaper+better option elsewhere? I'm a PCIE newbie so any general tips would be welcome.
As for SSDs, I know I need to look for a decent size and TLC+DRAM is okay. For my needs, 120-250GB would suffice. Is something like https://www.amazon.com/gp/product/B01K8A29BE/ a decent option? I might even go for two and run in a mirrored pool. Thoughts on that?
-
Seems to have worked?
Subject: Notice [TOWER] - Parity sync / Data rebuild finished (0 errors)
Description: Duration: 12 hours, 8 minutes. Average speed: 91.6 MB/s
Hopefully it doesn't regress in the next few days.
-
Here' what I've done:
1. Stopped the array and unassigned the parity drive.
2. Start the array, then stop and shutdown.
3. Checked disk connections again
4. Power on, add the same disk back as parity (no preclear - I forgot)
Now it's doing a parity sync. Hopefully this works.
If not, what else can I try? Repeat above with new SATA cable? New HDD?
-
Thanks. Send like rebuild is a pretty involved process - will try it out today.
Full diagnostics attached (apologies, first time reporting issues).
-
My party check came up fine. I have no idea if this is a real issue or not.
Thanks for the link, but I don't know if replacing disks is the first step i should follow here? How can I just get unraid to try the existing disk again?
I can perhaps try a new Sata cable but I would like to know how to clear the disabled bit without replacing the disk.
-
So got the "Parity disk in error state" notification today. At first, I don't see much in SMART (attach diagnostic).
I replugged the parity disk's SATA connection and spun back up the disk. I still see a red x next to the parity disk. I have started another parity check but I don't know if the red x means anything right now.
What do I do next?
-
Yeah I got Catalina 10.15.4 and no luck with TM over SMB.
-
-
Same as that. I have followed the guide, and I can mount the smb share but TM doesn't see it as a drive. I'm on Catalina.
-
Seems like it's still not working consistently:
Disk 1 is getting everything!
Here's my share settings again
When I `cp` something on to the share, it always goes to disk1. However when radarr copied completed downloads, it went to disk2. What the difference here? I'm using smb. Radarr runs in docker and mounts the same smb share that I cp files to.
EDIT: Nvm I think that's because the existing directories don't get moved, only new ones do. Still odd.
-
That makes a ton of sense - thanks for explaining!
-
I have 3 disks and a share with all 3 (called Media). My file structure is this:
/Media/Movies/Movie1/file.mkv
/Media/Movies/Movie2/file.mkv
/Media/Shows/Show1/S01/file.mkv
/Media/Shows/Show2/S01/file.mkv
You get the idea. Now, after reading a bunch, I figured a split level of 3 with high water allocation would spread out my media over disk. But in reality, this is what's happening:
Again, here's my share settings:
I have tried to experiment with split level of 2 as well but nothing seems to work. What do I need to do? Why does this not work?
NFS Stale File Handle (version 6.8.0) Tunable set to -1
in General Support
Posted
Turning off hard-links seems like a terrible idea for performance btw.