hiltera

Q: FINAL CUT SERVER, COMPRESSOR AND XSAN

Hi Everybody.

We’ve just implemented a Xsan Project in a TV Station and everything is doing well, but except the Final Cut Server.

We are using two Xserves (MDC1 and MDC2) and there are three MACPROs (Clients).

The MDC2 is running the Final Cut Server and what is occuring is described below:

When we upload the materials to Final Cut Server and when the material is being transcoding using the Compressor (from Xdcam to ProRes422 HQ) sometimes a crash occur.

What causes the crash is that the memory consumption of MDC2 is maximum and the final cut server doesn’t work.

I have 14GB of memory installed in the Xserve (MDC2) that is running the final cut server and we execute only three process at the same time and even with only three process the memory is totally used by them. So, the Xserve crashes every time this situation happens.

Do you have any idea of what can be occuring?
We already added to Xsan a third Xserve and installed the FCS in it, but the same problem occurred.

MACPRO WESTMARE 2.4 GHz MC561BZ/A

Posted on Mar 24, 2011 4:11 PM

Close

Q: FINAL CUT SERVER, COMPRESSOR AND XSAN

  • All replies
  • Helpful answers

  • by ignatiev,

    ignatiev ignatiev Mar 25, 2011 6:00 AM in response to hiltera
    Level 1 (0 points)
    Mar 25, 2011 6:00 AM in response to hiltera
    I think possible solution for you are:
    - Add memory to MDC2. As normal is 2 GB of RAM for 1 core at Compressor, and as i remember 2 GB of RAM for 10000 assets. And 2-4 GB of Ram for MACOS.
    - Increase the value of Retry Count of 1.5 times at Global Preference. Test this value. Maybe 2 times will will not be enough for you. So increase this value to more times.
    - Set Retry Timeout to 3-10 seconds.

    Good solution is move Compressor Cluster to standalone powerful computer.
  • by BenB,

    BenB BenB Mar 27, 2011 9:19 PM in response to ignatiev
    Level 6 (10,036 points)
    Audio
    Mar 27, 2011 9:19 PM in response to ignatiev
    It's also a VERY bad idea to run FCSvr on an MDC. An MDC should be dedicated, and not be a file server.
  • by ignatiev,

    ignatiev ignatiev Mar 27, 2011 11:05 PM in response to BenB
    Level 1 (0 points)
    Mar 27, 2011 11:05 PM in response to BenB
    You're absolutely right, and its documented fact. This is the standard recommended configuration of Apple.
    But as I understand it, @hiltera has 2 servers. One of them (MDC1) serves as the Main XSAN metadata controller. And in MDC2 was increased RAM to 14 GB, what would it launched Final Cut Server, and just in case its working Backup XSAN Metadata. I think this configuration of XSAN consist of 1 fiber switch. So if switch will be go down, MDC2 don't help in this situation. MDC2 can help @hiltera, when he want reboot MDC1. But in this case he can or stop Final Cut Server, unmount and stop XSAN volume. Or manual failover to MDC2, and will be used MDC2 as Backup XSAN Metadata Controller for a short time.
    mething wrong, if so, please correct me
  • by BenB,

    BenB BenB Mar 28, 2011 5:30 AM in response to ignatiev
    Level 6 (10,036 points)
    Audio
    Mar 28, 2011 5:30 AM in response to ignatiev
    Retry counts aren't going to use less memory. Cut back on the maximum number of jobs to run, keep it below 5, maybe 2 or 3. That and increasing RAM by doubling what he has is about all he can do.

    But an MDC uses a lot of resources, so does FCSvr. I'd personally monitor both MDC's for a few days and see how often the fail-over happens back and forth between the two.

    Now, in theory, what he wants to do will work, "technically". But I've seen it tried several times, and it always ends up with problems such as this. The reason to not run applications on MDC's is not about rebooting systems, it's about maxing out the MDC's resources.

    When the OP says it's maxing out memory, exactly what does he mean? And what is the storage drive setup?
  • by ignatiev,

    ignatiev ignatiev Mar 28, 2011 6:51 AM in response to BenB
    Level 1 (0 points)
    Mar 28, 2011 6:51 AM in response to BenB
    Maybe you're right.
    But I just checked, and at about this configuration, with one MDC and 6 clients (MacPro) I do not download RAM - 2.73 GB, CPU use - less than 5-10%. This data is up to 7 days.
    In my system i don't see failover from the start of using system. So i can't understand why you have failover...
    What is your configuration of XSAN?
  • by BenB,

    BenB BenB Mar 29, 2011 1:54 AM in response to ignatiev
    Level 6 (10,036 points)
    Audio
    Mar 29, 2011 1:54 AM in response to ignatiev
    Failover happens all the time, and you may never know it. It is why you have two MDC's in the first place.
  • by ignatiev,

    ignatiev ignatiev Mar 29, 2011 10:51 PM in response to BenB
    Level 1 (0 points)
    Mar 29, 2011 10:51 PM in response to BenB
    Sorry, but i don't understand, why failover happens in your system?