{“title”: “Who let the dogs out”}

Bulk delete assets (images and videos) from Immich

2025-08-27T00:00:00+00:00

TL;DR: I messed up while importing a big library into immich and need to roll this back somehow.

I’ve a semi large immich setup (5tb) with 2 users. I mistakenly uploaded around 500GB of assets into other user’s account instead of my own account due to my messup of the API keys through immich-go.

I was wondering if there is a clean approach to fix the ownership issue instead of deleting everything and starting over. Apparently not

I thought about doing something like this but I was not sure if the physical files can stay under /upload/user2_id while owner is user1_id and what that would mean for other sub components for immich:

UPDATE "asset" SET "ownerId" = 'user1\_id' WHERE "ownerId" = 'user2\_id' AND "createdAt" >= '2025-08-26';

Apparently, it’s somewhat possible but not everything will work without additional work in the db and possibly moving physical files according to the reddit Immich community, huge thanks!

Then I had to go for safer path by deleting the assets from the other account then reupload. Surprise, this is not easy either. Most of the info online is outdated, I don’t see “Recently uploaded” button anymore in the UI; Date based search is not giving the exact images I uploaded. So what can i do?

Let’s pull the asset ids from the db into a file since i know how to do it:

docker exec immich_postgres psql -U $DB_USERNAME -d $DB_NAME -c "SELECT * FROM assets WHERE \"ownerId\" = '$OWNER_ID' AND \"createdAt\" >= '2025-08-26';" > asset_ids_to_be_deleted.txt

Then I need some python power to read this file and call immich asset deletion endpoint through the api. ofc you can do that from curl too

Tricky part is setting the batch size to 1; I tried with multiple numbers, nothing worked even though api response is 200, the assets are not deleted.

SELECT count(*) FROM assets WHERE "ownerId" = '$OWNER_ID' AND "createdAt" >= '2025-08-26';

How to know if your laptop is bypassing the battery and using AC power directly

2025-07-29T00:00:00+00:00

I do lots of self hosting and some of my machines are old laptops with lids closed.

However, it’s not always possible to remove the battery from the laptop and use it with only AC power due to new slimmer laptops have embedded batteries.

If you have a battery discharging and charging 365/7/24 then you may start getting swollen battery and fire hazard arises. if you are a self hoster at home, noone wants that.

Here are some bash commands to know if your battery is being used or direct AC power is used.

sudo cat /sys/class/power_supply/AC0/online
# cat: /sys/class/power_supply/AC0/online: No such file or directory

sudo cat /sys/class/power_supply/BAT0/status
# cat: /sys/class/power_supply/BAT0/status: No such file or directory

ls /sys/class/power_supply/
# ADP1  BAT1

cat /sys/class/power_supply/ADP1/online
# 1

cat /sys/class/power_supply/BAT1/status
# Not charging

How to understand the discrepancies in the hard drive lifetime (smart farm logs)

2025-04-19T00:00:00+00:00

Preface: The discrepancy doesn’t mean that your hdd is bad. It’s an input for decision making. I use this to spot “new new” drives vs “refurbished but sold as new” drives especially on amazon.

I was extending my NAS disks and I had already got 2x 12TB Seagates from amazon.de. Since they have been behaving nicely for the last 6 months, I wanted to get 2 more. Then I saw a random review which mentions the farm logs showing different values than smart monitoring tools (s.m.a.r.t.) aka smartctl. I have never heard FARM logs before, apparently it stands for Field Accessible Reliability Metrics and useful additional metric for Seagate drives which is harder to reset than smart Power_On_Hours metric.

Guess i got scammed by this German Seagate hdd scammers who reset the s.m.a.r.t. information of the hdds and sell it as unused.

best way to check apparently is the discrepancy of "power on hours":

smart power on hours : 4083 (166 days)
seagate farm logs PoH: 28831 (1200 days) pic.twitter.com/ofY36fNDRy
— Hakan Uysal (@hakanu_) April 14, 2025

I researched on reddit and saw that there is a big market there: https://www.tomshardware.com/pc-components/hdds/german-seagate-customers-say-their-new-hard-drives-were-actually-used-resold-hdds-reportedly-used-for-tens-of-thousands-of-hours

However, noone is explaining the exact command to run. Here they are

sudo apt update
sudo apt install smartmontools 

# Usual smart logs for power on hours.
sudo smartctl -a  /dev/sdb | grep "Power_On_Hours"

# Only applicable for Seagate drives
# FARM Logs
sudo smartctl -l farm /dev/sdb | grep "Power on Hours"

If you drive is “new new”, you should get almost zero hours for both of these commands. However, in my case smartctl -a was giving almost zero but smartctl -l farm 28k hours which is totaling to ~1166 days. Obviously this doesn’t mean that the drive is bad or anything. Drive is manifactured probably around that time. However some refurbishing event happened and got it fixed and resold to me with clean S.M.A.R.T.

Solution of error getting credentials - err: exec: “docker-credential-desktop.exe”: executable file not found in $PATH, Docker on WSL Ubuntu

2024-10-05T00:00:00+00:00

Hit an error today, solution is very stupid, so I had to blog it here.

Error

$ docker compose up
[+] Running 0/0
 ⠋ scriberr Pulling                                                                                                                                                                                                                           0.0s
error getting credentials - err: exec: "docker-credential-desktop.exe": executable file not found in $PATH, out: ``

In ~/.docker/config.json change credsStore to credStore

$ cat ~/.docker/config.json
{
  "credStore": "desktop.exe"
}

Source

Mount exfat with read write access

2024-09-08T00:00:00+00:00

My external hdd i’m trying to mount to debian NAS server through usb is giving me permission denied access for write; it can only read which is not enough for my use case.

-o rw doesn’t work alone. Gemini keeps giving me only -o rw option for some reason.

sudo mount /dev/sdg1 /mnt/hdd_exfat/  -o  rw,uid=1000,gid=1000

If this still doesn’t work try this

# unmount first
sudo umount /dev/sdg1

# Change the ownership of the mounting point.
sudo chmown $USER:$USER /mnt/hdd_exfat

# Remount
sudo mount /dev/sdg1 /mnt/hdd_exfat/  -o  rw,uid=1000,gid=1000

Fixing always portrait apps in Huawei Tablet (Harmony OS)

2024-06-15T00:00:00+00:00

I got a cheap Huawei Tablet (Mate Pad Air) last year with very very good hardware, amazing screen, super snappy. All my tablet-worthy apps (photo/video editing and kids coloring pages apps) work great so it’s definitely a good deal for me. Among all the annoying things (no Google Play Store, not-as-apple-pencil quality stylus, inferior OS experience etc), the apps lock themselves into portrait mode regardless of the orientation of the tablet. Well it’s a tablet with keyboard which is always in the landscape mode. However, apps like Reddit, DJI Mimo etc always open themselves in portrait, forcing me to tilt the tablet with keyboard (or instead throw it to the wall).

I tried multiple things to solve the issue. Nothing helped, finally I stumbled upon a reddit post, I immediately opened it in my tablet, boom it got portrait-ed, I didn’t give up and read the post.

Apparently this is a common issue with wide devices, especially foldables and tablets. Since the manifacturers are lazy enough to fix these (small?) annoyance, Google introduced a method where the actual users can fix issue themselves, which is making the tablet ignore apps’ specification of rotation.

Steps are easy in the post, I need to connect via adb and then run this adb command. however nothing is easy.

Get the adb for windows, I’m nice enough so putting a link here to download and use adb without installing Android Studio and other gigabytes of software.
Now unzip and open terminal inside
./adb.exe devices => Nothing happens. Because you didn’t enable usb debugging in the tablet yet.
Go to tablet => Settings => System => Touch Software version multiple times to enable developer options.

Enable usb debugging, hopefully once you do that if you have data + charging usb-c cable, your tablet should prompt you if it should trust this computer. If this doesn’t show up, sorry, more steps are coming. Try to unplug and re-plug the cable, consider enabling usb debugging while charging too, maybe dangerous so don’t make it permanent.
Then if it still doesn’t work, try this, this did the trick for me: modify USB configuration from MTP (Media Transfer Protocol) to RNDIS (USB ethernet), then boom it asked if it should trust my pc. I approved and adb devices started to show my tablet.

And then finally we can run the magical command: .\adb.exe shell wm set-ignore-orientation-request -d 0 true
It should silently finish, no errors. Then fire up reddit app while your tablet is in landscape mode.

Looks ugly, doesn’t use all the estate but who cares, I don’t need to rotate the tablet physically anymore.

Enable x265 (HEVC) hardware decoding in Raspberry Pi 4

2024-02-20T00:00:00+00:00

My Emby Server struggles a lot during playing x265 formatted videos, it gets stutters every 3-4 seconds. Very annoying. I resolved like this:

Edit /boot/config.txt: sudo vim /boot/config.txt
Add this line to the end

dtoverlay=rpivid-v4l2

(More details about v4l2)

Save the file.
Reboot the pi: sudo reboot now

If this doesn’t solve your problem, check out the HEVC support in your browser. The client consuming emby content should also be compatible.

Solving “Error: TRIGGER_PAYLOAD_TOO_LARGE: This request would cause a function payload exceeding the maximum size allowed” in firebase realtime db

2023-12-24T00:00:00+00:00

I was using this piece and it stopped working in one night:

db.reference('foo/some_key').delete()

I wrote the code above in different variations, nothing helped. Writing it here so that I won’t forget. Eventually the solution was very easy, I had an inactive firebase function listening to firebase realtime database writes and I just deleted that function. This thing got resolved.

The inspo is from this old post:

https://stackoverflow.com/questions/50513374/firebase-realtime-database-currently-gives-trigger-payload-too-large-error

Hope saves someones less hours than me.

Solving “Invalid JWT: Token must be a short-lived token” error while using firebase realtime db on WSL

2023-12-24T00:00:00+00:00

I keep hitting this in my Windows laptop with WSL (Windows Subsystem for Linux).
This is the full error:

google.auth.exceptions.RefreshError: ('invalid_grant: Invalid JWT: Token must be a short-lived token (60 minutes) and in a reasonable timeframe. Check your iat and exp values in the JWT claim.', {'error': 'invalid_grant', 'error_description': 'Invalid JWT: Token must be a short-lived token (60 minutes) and in a reasonable timeframe. Check your iat and exp values in the JWT claim.'})

Check out the date: date => Probably will output something weird.

Sync it again (needs sudo):

sudo hwclock -s

Check out date date again: date => This time it should show correct timezone. if not, there are more commands to run I stole from internet.

sudo timedatectl set-timezone Europe/Amsterdam

date

# if this doesn't work
sudo apt-get update
sudo apt-get install -y ntpdate

sudo ntpdate pool.ntp.org

Fixing mongodb start error code 14/n/a

2023-08-15T00:00:00+00:00

Stupid errors today.

My local mongodb which has been running for multiple weeks decided not to start.

First I need to remove MONGODB_CONFIG_OVERRIDE_NOFORK env variable.

sudo vim /usr/lib/systemd/system/mongod.service
# Convert this line:
# Environment="MONGODB_CONFIG_OVERRIDE_NOFORK=1"
# To This line:
# Environment="MONGODB_CONFIG_OVERRIDE_NOFORK=0"

Then it still refuse to up itself, but this time no errors:

sudo systemctl start mongod

x:~$ sudo systemctl status mongod

× mongod.service - MongoDB Database Server
     Loaded: loaded (/lib/systemd/system/mongod.service; enabled; vendor preset: enabled)
     Active: failed (Result: exit-code) since Mon 2023-08-14 20:33:54 CEST; 1s ago
       Docs: https://docs.mongodb.org/manual
    Process: 1501 ExecStart=/usr/bin/mongod --config /etc/mongod.conf (code=exited, status=14)
   Main PID: 1501 (code=exited, status=14)

Aug 14 20:33:54 x systemd[1]: Started MongoDB Database Server.
Aug 14 20:33:54 x systemd[1]: mongod.service: Main process exited, code=exited, status=14/n/a
Aug 14 20:33:54 x systemd[1]: mongod.service: Failed with result 'exit-code'.

Apparently it’s a permission issue:

sudo chown -R mongodb:mongodb /var/lib/mongodb
sudo chown mongodb:mongodb /tmp/mongodb-27017.sock
sudo service mongod restart