When recording continuously with motion trigger recording enabled, I lose 10-60 frames at the beginning and end of the motion event.
I am running the latest stable BI5, with four cameras currently, two amcrest, two unifi, though it seems to happen with any manufacturer/model I try.
I have my rig set to record continuously, with BI handling motion detection just for convenience and ease of locating events (the unifis suck hard at it, & if I'm going to have to use software anyway, I'm going to use it for all of them). They are all recording to an external thunderbolt WD purple. The cams and BI server are all on the same switch and subnet so that they can talk at wirespeed, no bottlenecks.
I have tried changing it so that camera motion events, and even an IR sensor, were the trigger, but I still lost frames. I increased the buffer time, which increased the length of the recording of the motion event, which was somewhat helpful, but I still lost frames at the beginning and end of the event. Even when I export the continuous recording of the 5 minute period in which the motion event took place, which is theoretically a separate file from the motion event itself, the dropped frames persist.
Now that I have completely disabled alerts, things appear to be running smoothly - I can't find any new instances of dropped frames. Since that's the only significant change I made, it would seem to rule out network, hardware, and camera issues. I also briefly considered that I might be hitting the I/O bottleneck for the external drive, but even during peak activity it only seems to be using about 15% of I/O capacity, and I make sure to leave at least 250GB free just in case.
Have I discovered a known limitation in the technology, a BI5 bug, or am I doing something wrong?
I am running the latest stable BI5, with four cameras currently, two amcrest, two unifi, though it seems to happen with any manufacturer/model I try.
I have my rig set to record continuously, with BI handling motion detection just for convenience and ease of locating events (the unifis suck hard at it, & if I'm going to have to use software anyway, I'm going to use it for all of them). They are all recording to an external thunderbolt WD purple. The cams and BI server are all on the same switch and subnet so that they can talk at wirespeed, no bottlenecks.
I have tried changing it so that camera motion events, and even an IR sensor, were the trigger, but I still lost frames. I increased the buffer time, which increased the length of the recording of the motion event, which was somewhat helpful, but I still lost frames at the beginning and end of the event. Even when I export the continuous recording of the 5 minute period in which the motion event took place, which is theoretically a separate file from the motion event itself, the dropped frames persist.
Now that I have completely disabled alerts, things appear to be running smoothly - I can't find any new instances of dropped frames. Since that's the only significant change I made, it would seem to rule out network, hardware, and camera issues. I also briefly considered that I might be hitting the I/O bottleneck for the external drive, but even during peak activity it only seems to be using about 15% of I/O capacity, and I make sure to leave at least 250GB free just in case.
Have I discovered a known limitation in the technology, a BI5 bug, or am I doing something wrong?