Giters

awesometic / docker-h5ai

Lightweight h5ai 0.30.0 container with Nginx 1.21 & PHP 8 based on Alpine Linux.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

webm thumbnail failure

breadcat opened this issue · comments

commented

Hello there, I've recently replaced corfr/h5ai with your container but have noticed that I no longer get thumbnails for webm files.
When I click them to preview the file fullscreen, it will work fine (at least in my browser, Firefox).

The error I'm getting via docker logs when requested is as follows:

172.18.0.12 - - [11/Nov/2021:16:39:34 +0000] "GET directory/file.webm HTTP/1.1" 206 131147 "https://domain/directory/" "Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:94.0) Gecko/20100101 Firefox/94.0" "myip"
2021/11/11 16:43:17 [error] 41#41: *87 FastCGI sent in stderr: "PHP message: PHP Fatal error:  Allowed memory size of 134217728 bytes exhausted (tried to allocate 36864 bytes) in /config/h5ai/_h5ai/private/php/ext/class-thumb.php on line 148" while reading response header from upstream, client: 172.18.0.12, server: _, request: "POST /directory/ HTTP/1.1", upstream: "fastcgi://127.0.0.1:9000", host: "domain", referrer: "https://domain/directory/"
commented

Hello. Sorry for the late reply.

Can you attach a sample webm file?
I can see a thumbnail of some sample webm files I downloaded from the Internet.

Screenshot_20211129_125327

commented

Thanks for your reply, I've attached an affected file..

It's strange, I've tried using two example files from your links above and both of these work absolutely fine.

I originally converted these files from gifs to webm to save space using ffmpeg, I have also issued file *.webm and all of them identify as WebM files. I assume this conversion process is where the issue will lay.

commented

I have looked into that, but I don't know what is the actual differences between the sample files from those sites and the one ffmpeg generated.

I converted a sample video file to a webm file, on my side, but these files are not providing a thumbnail either.
Here's what I tried to.

// VP8
ffmpeg -i /h5ai/sample.h264 -c:v libvpx -crf 30 -b:v 0 /h5ai/temp/vp8.webm

// VP9
ffmpeg -i /h5ai/sample.h264 -c:v libvpx-vp9 -crf 30 -b:v 0 /h5ai/temp/vp9.webm

Did you report this issue to the h5ai developer?

commented

Thanks for your reply, sorry for the long response.
I think it's safe to say that this isn't an issue with your container with the sample files working fine. I'll raise this either with ffmpeg or h5ai.

Thanks for your help.