-
Posts
479 -
Joined
-
Last visited
Content Type
Profiles
Forums
Downloads
Store
Gallery
Bug Reports
Documentation
Landing
Posts posted by CS01-HS
-
-
Makes sense. I thought maybe I misunderstood AVR or my unit had a substandard implementation. Thanks. Sorry OP for the tangent.
-
1 hour ago, Hoopster said:
The most important consideration is a UPS that supports Automatic Voltage Regulation (AVR).
My Cyberpower's AVR only kicks in when voltage drops down from 120V to 104V. That rarely happens so why's it important the drop's supplemented by circuitry rather than battery? The more common case (total power loss) is handled by battery entirely.
-
5 minutes ago, ken-ji said:
Odd. I've never seen any transcoding files - I'm transcoding to RAM vi a mounted directory under /tmp - unless I look while someones actively playing something. So I've never this issue.
Same. In my case multiple simultaneous transcodes maxed out the RAM drive.
-
40 minutes ago, StevenD said:
Thanks! Funny enough, I couldnt figure out why my cache drive was almost full this morning. I found a 445GB .ts file in the Emby appdata/transocde dir.
I've never seen a file that size. Mine are few MB each. The large size suggests it's being continuously written to in which case my script deleting it may cause whatever's writing it to fail...
-
Emby doesn't intelligently manage temporary transcoding files and can fill your drive in some scenarios.
This is especially a problem if you're transcoding to a RAM drive with limited space as described here (the specified command is added to template's Extra Parameters.)
Update: I've switched from tmpfs (outlined here) to /dev/shm
I solved this with a helper script stored on a share accessible to the Emby container and called from within it.
Create it as transcoding-temp-fix.sh and run chmod a+x on it.
Here's the script:
#!/bin/sh # User Settings TRANSCODE_DIR="/transcode/transcoding-temp" PERCENT_LIMIT=70 BATCH_SIZE=10 if [ -d "${TRANSCODE_DIR}" ]; then percent_full=`df "${TRANSCODE_DIR}" | awk '{print $5}' | tr -dc '0-9'` echo "Directory limit: ${PERCENT_LIMIT}%" echo "Directory utilization: ${percent_full}%" while [ $percent_full -gt $PERCENT_LIMIT ]; do echo "(${percent_full}%) exceeds limit (${PERCENT_LIMIT}%), deleting oldest (${BATCH_SIZE}) files" # DEBUG: Uncomment to print list of deleted files #ls ${TRANSCODE_DIR}/*.ts -1t | tail -${BATCH_SIZE} | xargs ls -lh ls ${TRANSCODE_DIR}/*.ts -1t | tail -${BATCH_SIZE} | xargs rm percent_full=`df "${TRANSCODE_DIR}" | awk '{print $5}' | tr -dc '0-9'` done else echo "${TRANSCODE_DIR} (TRANSCODE_DIR): directory doesn't exist" fi
This assumes you're mounting the RAM drive at /transcode (as described in the linked post) which you've set as the Transcoding temporary path in Emby:
The script deletes the oldest ts files in batches of 10 (BATCH_SIZE) until it reaches a safe usage percent (PERCENT_LIMIT).
I call it every 30s with the following entry in the template's Post Arguments
&& docker exec EmbyServer sh -c 'watch -n30 "/system-share/transcoding-temp-fix.sh" 2>/dev/null > /transcode/transcoding-temp-fix.log &'
NOTE: The full host path to the script is /mnt/cache/system/emby/transcoding-temp-fix.sh where system-share is a mapped path variable I've added to the template:
Its execution can be monitored with the following command from the unraid console:
docker exec EmbyServer sh -c 'tail -f /transcode/transcoding-temp-fix.log'
It works well so far with standard transcoding but I don't use batch conversions or live TV so YMMV.
NOTE: I'm piping results to rm so use at your own risk (and feel free to post improvements.)
You may want to test the commands individually before running the full script.
-
1
-
-
Any plans to include iHD for QSV in addition to i965?
Otherwise does anyone know of a docker that will convert to HEVC with hardware support for encode AND decode? Handbrake works but no HW decode support.
-
I removed the Trim plugin (on RC2) but now Fix Common problems is warning me that it's not installed - should I ignore it?
-
Updated with miscellaneous reliability and display tweaks and switched shutdown trigger from battery level to time-on-battery.
-
Anyone encoding h265 with quick sync mind sharing their settings?
Mine (defaults except for 192kbps audio) practically doubled the size of the h264s it converted.
-
13 hours ago, JimmyGerms said:
Thanks for the tutorial. It helped me get to where I needed to go!
Glad it was helpful.
Actually I've updated it with misc reliability tweaks and a shift from battery level-based shutdown to timer-based shutdown. I was holding off until I had a few natural power outage/shutdowns to test with but I'm reasonably confident it's an improvement.
-
On 1/8/2021 at 12:33 AM, JimmyGerms said:
NUT will reset ups.delay.shutdown and ups.delay.start to their default values of 20 and 30 blowing the whole setup randomly.
Are you sure it's random? If e.g. it's reset on startup you can save the good version to your boot drive and copy it over in config/go (which I'm pretty sure runs after plugins are installed)
-
As long as you're happy with 6.8.3.
6.9.0 dropped AFP support.
-
Wow that's comprehensive.
1 hour ago, theruck said:conclusion: AFP protocol performs much better than SMB protocol on a MAC. And yes i did all the recomended tweaks on MAC side.
It's unfortunate Apple chose to deprecate it.
I was surprised to see in your tests with small files that Windows wasn't much better than Mac. I assumed Mac was the biggest culprit but this suggests it's SMB itself or unraid's implementation. I guess to narrow that down you could benchmark a non-unRAID SMB share.
-
On 1/20/2021 at 6:09 AM, theruck said:
there are plenty of recomendations out there for SMB tunings. none of them seems to work for me. my sparsebundle performance over SMB is so terrible that it is unusable for timemachine.
I haven't tested enough to gauge the relative effectiveness of each of these tweaks (some may even be inadvisable) but overall my Time Machine backups are working "reasonably" in 6.9.0-RC2
In Samba extra configuration:
[global] #Fix for 6.9.0-rc2 Mac client search spotlight backend = tracker # speed tweak nt acl support = No # tweaks from https://wiki.samba.org/index.php/Configure_Samba_to_Work_Better_with_Mac_OS_X min protocol = SMB2 fruit:metadata = stream fruit:model = MacSamba fruit:posix_rename = yes fruit:veto_appledouble = no fruit:wipe_intentionally_left_blank_rfork = yes fruit:delete_empty_adfiles = yes #Custom include = /boot/config/smb-custom.conf
Where smb-custom.conf looks like this:
[Time Machine] path = /mnt/fast/Time-Machine comment = browseable = no writeable = no read list = write list = backups valid users = backups vfs objects = catia fruit streams_xattr fruit:time machine = yes fruit:time machine max size = 552000M case sensitive = auto preserve case = yes short preserve case = yes spotlight = yes
And fast is a cache-only NOCOW BTRFS RAID 5 pool to which this custom conf allows direct access (bypassing shfs.)
More details on that here:
Finally, I wiped my previous sparsebundle and started fresh in Big Sur to take advantage of the new APFS formatting.
NOTE: The user "backups" in smb-custom.conf is one I created just for time machine (for security reasons from an older guide.)
-
On 1/20/2021 at 10:31 AM, ken-ji said:
Looks like a bug with the shfs used by Unraid.
Based on your own testing or what I reported?
Every container with root access seems like a significant security risk.I thought this would get more interest. Am I misunderstanding?
-
10 hours ago, discojon said:
I've resorted to that. It just takes a while to extract a 35Gb rar file.
-
22 minutes ago, itimpi said:
I would like confirmation it has done it for others.
Confirmed.
-
I'm having the same problem with "missing csrf_token."
I tried a server restart to fix it, didn't help.
-
Update:
As of Emby 4.6.* (currently in beta) which includes an iHD driver with fixes for Gemini Lake it's no longer necessary to disable iHD.
-
This is worse than I thought.
I created a directory in the unraid console owned by root and not readable, writable or executable by anyone else:
mkdir /mnt/user/Download/TV/y chmod go-wxr /mnt/user/Download/TV/y
drwx------ 1 root root 20 Jan 11 18:49 y/
And apparently any user in my sonarr container (not set as privileged) can write to and delete from it freely.
This can't be intended behavior, can it?
-
18 minutes ago, itimpi said:
Have you checked the permissions on the containing directory? I think that ‘wx’ permissions allows deleting files within the directory.
Thanks. A slight variation on your suggestion, I re-ran the test but with x as directory (rather than file) - same problem:
docker exec -u 99 -it EmbyServer sh
/ $ mkdir /media/TV/x / $ chmod go-wx /media/TV/x / $ ls -l /media/TV/ | grep 'x' drwxr--r-- 1 99 root 0 Jan 11 12:09 x
So directory x should only be writable by nobody
Let's see if that's true:
docker exec -u 1006 -it sonarr sh
$ whoami abc $ grep abc /etc/passwd abc:x:1006:100::/config:/bin/false $ ls -l /tv/ | grep x drwxr--r-- 1 99 root 0 Jan 11 07:09 x
$ echo 'dodeedododeedo' > /tv/x/file.txt $ ls -l /tv/x/file.txt -rw-r--r-- 1 abc users 15 Jan 11 07:17 /tv/x/file.txt $
Nope!
-
I created an unRAID user limited (id 1006) which I'm trying to use to limit write permission in docker containers that support PUID.
Can anyone explain the following behavior?
Launch EmbyServer shell with user nobody
# docker exec -u 99 -it EmbyServer sh
Create a file writable only by nobody
/ $ touch /media/TV/x / $ chmod go-w /media/TV/x / $ ls -l /media/TV/x -rw-r--r-- 1 99 root 0 Jan 11 11:35 /media/TV/x
Launch sonarr shell with user limited
# docker exec -u 1006 -it sonarr sh
Confirm I'm limited (id 1006):
$ whoami abc $ grep abc /etc/passwd abc:x:1006:100::/config:/bin/false
Check permissions on the newly-created file:
(NOTE: /media/TV in EmbyServer and /tv in sonarr map to the same host path.)
$ ls -l /tv/x -rw-r--r-- 1 99 root 0 Jan 11 06:35 /tv/x
So the file should not be writable by current user abc - and yet:
$ rm /tv/x $ ls -l /tv/x ls: cannot access '/tv/x': No such file or directory
abc can delete it.
What's going on?
-
On 12/6/2020 at 8:47 AM, CS01-HS said:
I managed to cause the same error (and freeze my server) playing around with intel_gpu_top in the intel-gpu-tools container so my problem's at a lower level, your container's fine.
After several freezes (which caused unclean shutdowns) in Handbrake using the hardware encoder and also with monitoring, and random corrupted encodes, I got my machine stable with the following changes:
- Added intel_iommu=on,igfx_off to syslinux config (this may be optional)
- Added a dummy HDMI plug to my headless server (j5005)
It's been stable now for several days despite continuous hardware encoding in Handbrake with no corruption.
-
Are you sure it won't work with the built-in "UPS Settings" ? Support seems to vary model-to-model and mine's a low-end unit.
Otherwise I can't help much with customization of the NUT plugin. The reason I connected it to the Pi was to allow me to tweak it. Maybe post to that thread?
Spin down command for pool drives and unassigned devices?
in General Support
Posted
hdparm seems to work but is there a better way (that might update the WebGUI's activity "LED") ?