On 06/14/2012 07:53 PM, Greg Logan wrote:
On 12-06-14 01:32 AM, Kristof Keppens wrote:
Hi List,
After a while we decided to start testing the epiphan devices again with
the latest firmware from epiphan's site ( 2.3.1e1 ) . Our results so far
aren't really promising. The first day of recordings went relatively
successful, 3 captures, around 5 hours together was ingested successful.
We did notice than when setting the stream setup to picture in picture
the ingested material is also picture in picture and not 2 separate
streams. For the second day of testing we changed the setup to 2
streams, however none of those recordings made it to the server, idem to
the recordings for the third day.
I have seen similar behaviour with my devices when my *cores* ran out of
space. Do you have free space remaining for your cores? Other than
that the devices with that firmware have performed very well in my testing.
My matterhorn core server has more than enough free space left, so that
shouldn't be the issue.
I have in the capturing overview in matterhorn 2 recordings from the
second day ( the third recording that day doesn't even show there ), one
stuck in capturing and one stuck in sending recording to processing. For
the third day only 1 recording made it to the overview, also stuck in
sending recording to processing.
In the logs of the epiphan device I notice a message for the recording
stuck in capturing that says unable to save recording ( or something
similar ) and for the first recording of the third day a message keeps
appearing that says trying to ingest, and zipping to /u/data/, this
repeats for a while and at last it will disappear only to start all over
again.
Is this testing with a 1.2 core, or 1.3? The 1.3 core (and newer) does
not require the zipping step, and this saves half of the storage space
on the device that you would otherwise need for zipping the files up.
It sounds to me like the internal storage is full, and this is
preventing the zip-based ingest from functioning properly.
Unfortunately I don't think there's a user-accessible way to clear the
disk right now. Is the device on a public IP? I can log in and wipe
out whatever is on it for you.
I'm testing this on a 1.2 core at the moment, we are planning upgrading
to 1.3.1 for the next semester. If I understand correctly, even with the
1.3.1 core this problem may occur, it may just take a bit longer before
it happens? We normally have captures of around 2 hours, 2 or 3 in a row
with a short break between them. Will the device with an 1.3 core be
quick enough to keep recording, ingesting and cleaning this number of
captures before filling up again?
The test recordings are scheduled as follows :
10:00 -> 11:45
12:00 -> 13:55
14:00 -> 14:45
And this 5 days in a row. Could it be possible that after the first 3
recordings ( that succeeded ) the disk is full and therefor creating
this problem or are we doing something else wrong. Another question is,
Yes, this is quite possible. In my testing I have three devices each
capturing a 2:50 minute capture, and I can only have each device capture
three times before I fill up my cores. It's entirely possible that
capturing a long capture on the device can also take up enough space
that the device will be unable to ingest properly with a 1.2 core,
however there just isn't anything I can do since the fix for this
changes the ingest API.
is there already a way to login to the epiphan device and directly see
the current and past logs there cause the device log page has it's use
but it would be useful to be able to see all logs to pinpoint the issue.
Not currently. The device doesn't keep much in the way of logs
unfortunately, the memory precludes this. I'm hoping to get some remote
logging stuff done in the near future which would alleviate this issue.
G
Great, some remote logging would be a great help! We will wait a bit
until we moved to 1.3 servers and test the device again then.
Kristof
Thanks
Kristof Keppens
_______________________________________________
Matterhorn-users mailing list
[email protected]
http://lists.opencastproject.org/mailman/listinfo/matterhorn-users
_______________________________________________
Matterhorn-users mailing list
[email protected]
http://lists.opencastproject.org/mailman/listinfo/matterhorn-users
_______________________________________________
Matterhorn-users mailing list
[email protected]
http://lists.opencastproject.org/mailman/listinfo/matterhorn-users
