overseerr: Overseerr loading times are excrutiatingly slow

Description

It takes more than a few minutes to load any page on Overseerr. The screen just stays in Loading mode. Even the Settings page which is text only can take multiple minutes to load.

Version

1.33.0

Steps to Reproduce

  1. Go to http://OVER.SEE.RR.IP:PORT/requests
  2. Wait for page to load
  3. Wait some more
  4. Refresh hoping page load will be faster
  5. Repeat from step 2

Screenshots

Requests page after waiting for 4 minutesScreen Shot 2023-06-18 at 10 21 27 AM

Settings page after waiting for 30 secondsScreen Shot 2023-06-18 at 10 21 48 AM

Logs

No response

Platform

desktop

Device

Macbook Pro

Operating System

MacOS 11.7.7

Browser

Firefox 114.0

Additional Context

Overseerr is running on Synology NAS DS918+ using linuxserver.io docker image.

Code of Conduct

  • I agree to follow Overseerr’s Code of Conduct

About this issue

  • Original URL
  • State: open
  • Created 3 years ago
  • Reactions: 13
  • Comments: 58

Most upvoted comments

👋 hiya all, I took a bit of a deep dive on this and was able to solve it - at least for my use case (and explain why many folks can’t reproduce the issue!). it’s possible there are more reasons for slowdowns, but this might help some people.

tldr: it’s DNS. specifically, it’s DNS within Docker. tldr solution: edit your /etc/docker/daemon.json file to have your network’s DNS (or use - dns: <ip address> in docker-compose.yml etc

basically, every request within Overseerr was hitting Docker’s internal DNS, timing out after 4 seconds, then moving on to actual DNS servers. that’s why some people using bridge/host networking saw improvements, or non-Docker dev environments couldn’t reproduce it.

anyway, the deep dive, if anyone wants to see if it’s relevant for their setup:

[click me] reproduction steps/walkthrough I started by identifying a single slow call (`https://overseerr.mydomain.tld/api/v1/discover/watchlist`), commented out _most_ of the code within the container, until it was pretty much only doing a single `axios.get()` (i.e. no caching, no gzip, etc) to the Plex metadata server. it was still taking ~5 seconds.

(I figured the Plex metadata API was doing some rate limiting, however running via curl locally was fast every time.)

I noticed the container had wget available, so I ran the command within the container. it still took around 5 seconds. weird!

note: at this point we know it’s not an Overseerr problem, and it’s why people can’t reproduce it in local (non-Docker) dev environments etc.

the container has nslookup, so I ran it within the container:

# time nslookup metadata.provider.plex.tv
Server:         127.0.0.11
Address:        127.0.0.11:53

[... response data ...]

real    0m 4.00s   <- 4 seconds!
user    0m 0.00s
sys     0m 0.00s

and then by specifying my actual DNS server at home:

/app # time nslookup metadata.provider.plex.tv 192.168.x.x
Server:         192.168.x.x
Address:        192.168.x.x:53

[... response data ...]

real    0m 0.00s   <- instant
user    0m 0.00s
sys     0m 0.00s

(note: the slow lookup was repeatable, not just fast the 2nd time because of DNS caching)

I found a SO post (here) that described the exact 4 second slowdown, so we know it’s a Docker problem (or feature, I guess).

so I updated /etc/docker/daemon.json, restarted Docker, and re-ran the nslookup steps: instant responses for both. and Overseerr runs virtually instantly now.

note: this will almost certainly break setups that rely on using container names to lookup IPs. but at least Overseerr will be fast 😉

fwiw, I found a couple of other points that might improve some Overseerr performance in specific ways, but didn’t seem to impact my setup much:

  • node-cache has a config option useClones which defaults to true, and can be expensive on large cache values. some folks have reported improvements setting that to false
  • the newer version of connect-typeorm has a fix to prevent writing the expiredAt value on every request (ref), which might unlock SQLite reads on busy servers

I am experiencing the same issue. Extremely long load times on every screen, it started after upgrading to 1.33. I am running hotio docker image.

There doesnt appear to be anything in the logs that hints at why the loading is so slow.

I had this slowness too and for me what fixed it was removing the network_mode: bridge in my docker compose. Immediatly became super fast.

I have the same problem. Since i put overseer behind traefik everything is slow as hell. 20 Second load times for the first /api/v1/search?page=1&query=test request.

Confirmed @notjosh findings. Setting DNS internal to the container resolved my painfully slow overseer issues. Thanks @notjosh !

There seems to be a dozen posts about this on reddit and clearly a ton of people are having this issues and found their way to this thread. I am also using Cloudflare to access, so it seems to be DNS related. But why? No other web app I’m hosting has this issue.

  • Synology ds1621+
  • Docker (was) on HDD

Tried quite a bit to solve it, including debugging the app code in the container (+DNS/network) with no obvious issues on the app side…

Decided to add an nvme pool and move docker there, on the way noticed that here I have a checkmark here: Control Panel > Shared Folder > Docker [edit]

Unchecking it before moving docker folder to the nvme pool solved the issue for me.

I suspect the “data integrity” thing was the root - Overseer is pretty aggressive with file operations (cache) and Synology with counting checksums on it 😃

UPD: in theory, a simple ways to check: copy existing data folder to a different location where the “data integrity” is off and mount it.

FWIW — in my case, it turned out that setting DNS to 1.1.1.1 and 1.0.0.1 in my docker image was the culprit of the slowness. I had accepted that it wasn’t “great” for a while — but then pulling up overseerr logs it was logging a number of getaddrinfo EAI_AGAIN errors.

I removed that config, blew away the old overseerr db/cache (kept the settings.json), and recreated the docker image and it’s blazing fast now. I could’ve sworn I tested it without that DNS config, but perhaps that was on my old host (moved from a synology nas to a mac mini — but it was still slow and I gave up for a while).

Here’s my current docker config:

  overseerr:
    image: linuxserver/overseerr:latest
    container_name: overseerr
    environment:
      - PUID=501
      - PGID=20
      - TZ=America/Los_Angeles
    volumes:
      - $HOME/docker/overseerr/config:/config
    restart: unless-stopped
    network_mode: bridge
    depends_on:
      - sonarr
      - radarr
    ports:
      - "5055:5055"

Hope this helps someone else! It’s been bothering me nuts for months and it seems to be finally fast AF

I just added the extra docker run arg that @Schu- suggested. It immediately fixed my issue!

I am running my network DNS through PiHole. PiHole resolves requests using 1.1.1.1 already so maybe I have something misconfigured but not sure where to start looking. I’m just glad this has resolved it for the time being.

Hello All,

I just wanted to comment on this since I was able to figure out and fix my issue. After a bit more scouring the web, I was able to find a reddit post talking about slowness. https://www.reddit.com/r/Overseerr/comments/sirat6/discover_page_takes_3060_seconds_to_load/. This post talks about adding a DNS command to docker run --dns 1.1.1.1. I made this change on my system and Overseerr worked 100% immediately. This made me question my own DNS issue and did some digging. On my side I found a misconfigured ACL rule to allow my container to reach my DNS server. Once I fixed that all was good on my side.

I would recommend anyone having this issue to review the post about and possibly add that docker run command.

Thanks!