CasparCG Server 2.3 has a been plagued by a bug causing a memory and CPU leak when doing LOAD and LOADBG in the FFmpeg Producer. This bug has been tracked in this issue: https://github.com/CasparCG/server/issues/1356
There’s now a beta version with a fix available here:
Please test and report success / problems in the GitHub issue!
We’re trying to see if this impacts the Linux version as well.
If you have 2.3.1 LTS on Linux, please try sending multiple LOAD commands per second to the FFmpeg Producer while monitoring the CPU and RAM, and please report back!
Is this including the Flash EOL hot fix?
Yes, it includes all the changes in 2.3.1.
It doesn’t have the flash config tool or pdf in the zip, ill try and remember to do so for the final release
Thank you for the release, it’s very promising.
I’ve been trying to update some servers from 2.0.7 to 2.3.x but I cannot find any information about changes in the “info template” command to retrieve the fields form flash templates.
The message it delivers in 2.3.x (all three releases) is “Info OK” and that’s it, no fields at all.
I’m testing with the template called “caspar_text” included in some releases and it works well in 2.0.7 and 2.0.7.2
Does anyone use this command and knows how to get it to work in the new servers?
Thanks,
That has been removed from versions above 2.0.7 for a reason I also do not understand. Some performance issues. If I remember right.
Oh I didn’t see that in the change log. It’s a shame, we’ll have to get them some other way.
Thanks.
Here’s a discussion about the INFO command. Feel free to clarify what you’d like to see there: AMCP INFO regression · Issue #1151 · CasparCG/server · GitHub
The appropriate place for INFO TEMPLATE to be implemented now is in the media-scanner. There is an issue for it but nothing has been done INFO TEMPLATE · Issue #38 · CasparCG/media-scanner · GitHub
It could be hooked up to the same amcp command once something is implemented for it.
The code in 2.0.7 doesn’t need a flash player to do the check, it does a small bit of manual parsing of the template file, so it should be fairly straightforward to implement
I actually posted there last year and totally forgot about it.
If it gets into the scanner it would be great.