FFMPEG Crash! in FFAStrans Manager

[vijm1972]

Hello Steipal,

Need help on "FFMPEG Crash" issue.

I have created watchfolder base workflow on shared location and configured the same on three system. I had also configured work media cache on shared location on all the three system. My source files are in SD 50 Mbps MXF files and I am converting it to SD 15 Mbps MXF output. When I use the workflow only on one system I am not getting any error but if I configured it on multiple system I am getting error "FFMPEG Crash". I am using 12 core CPU system and only 2 cpu currently active. I had also tried to active only 1 cpu but still I am getting same error. My CPU utilization is above 50% @ 1core used.

Please let me know if I am missing something.

Thanks
Vijay

[admin]

Hi Vijay,

In order to get full farming functionality with FFAStrans you need to run it from one single share.
Example:
You have 3 separate systems connected through standard SMB networking. Place your FFAStrans folder on a shared path like f.ex. "\\system1\FFAStrans" And make sure ALL 3 systems with read/write-access run FFAStrans from that path. Now you need to set "Work media cache" to a shared folder also with read/write-access for all 3 systems and select the "Shared" option in FFAStrans configuration. Can you please verify this setup?

-steipal

[vijm1972]

Steipal I am using Isilon storage and all the three system accessing the application using one shared location. All the three system having full access of all the paths (shared media cache and inbox/outbox location). If I ran the application as standalone with same location it's worked properly but when I used in shared mode FFMPEG get crashed sometimes.

[admin]

Can you please e-mail your workflows and ffastrans.ini to me at post@ffastrans.com? I will try to recreate the issue.

-steipal

[vijm1972]

I had emailed you require details.

[vijm1972]

Steipal have you able to recreate the issue at your end?

[admin]

Yes and no. I have been able to get an FFmpeg-crash both on standalone and farming setups. Nothing so far pinpoints to farming causing more crash than standalone and I've tested two different farms. Either way, FFAStrans relies on the stability of ffmpeg and if ffmpeg crashes, it's not obvious that FFAStrans is the cause. I will however continue to investigate the issue.

To further help me out, can you please send your custom ffmpeg from the main FFAStrans folder\Processors\.enc_customff? Also, if it's possible; a small sample file of the kind of files you're transcoding.

-steipal

[vijm1972]

In my case I have not observed crash on standalone mode using same paths (Input/Output and work media cache). I am running the application from shared location. But when I used same workflow in farming mode FFMPEG get crash.

[vijm1972]

Hi Steipal,

Still I am facing same issue while processing jobs in FFAStrans manager mode. I have added around 20 servers and all of them are accessing FFAStrans from one shared location. Sometime FFAStrans get crash due to unknown error showing message "FFAStrans processor manager has stopped working". I have sent you custom FFMPEG preset.

Thanks
Vijay

[admin]

Hi again Vijay,

I'm sorry about the delayed feedback but I've just come back from my vacation.

But, wow! You've added 20 servers??? That's way more than I would imagine anyone would do but I'm very happy you put FFAStrans through that ordeal!

I think I can say with pretty high confidence that the crash has nothing to do with your custom FFmpeg preset. The program you see crashing does not know anything about your workflow. It's a "dumb" job distributor.

Please tell me:
1. Does the crash happen on the same servers every time?
2. What kind of network server holds the FFAStrans executable?
3. Does the crash happen only after the servers have been active with FFAStrans for a while?

-steipal