mocord vs bulk frames, creates events of 1sec to 30mn+

[Pedulla]

... but generally isn't there a setting to have ZM rewind and show a fair amount of frame before an alarm? Or once something is tagged as bulk, there is no going back and allowing to have ZM unbulk it?

There is the pre-event buffer. It's basically the number of frames to keep pre-event frame. It is a frame count, so the frame rate of your camera will determine how much "time" is recorded pre-event.

Screenshot at 2018-04-20 13-48-54.png (20.07 KiB) Viewed 8217 times

[marcmerlin]

Thanks Pedulla, I forgot where it was. Indeed that's what I was referring to.
This does not work with mocord and bulk frames either.
I get no pre-alarm frames in many cases (if not all) and my prevent image count is 25 (which I think is the default)

[jbg70]

I think pre-event buffer in mocord mode is a non-sense, because in you are recording all the frames.
If bulk frame would work correctly, you can browse all the frames starting from the alarmed frame going "Prev" and "Next".

Even if all the images are into the filesystem, at this moment, you can start "event investigation", displaying the alarmed image and you can view "Prev" image only if you are under frame#100.
You can view the "Next" images until you arrive at the next bulk frame.
This did not happened in version 1.29 where you can view all the frames.

@marcmerlin, if you do not have high load, and you are experiencing frame loss, in particular way when an alarm occurs, I solved the problem adding the following lines into my.cnf:

Code: Select all

[mysqld]

innodb_read_io_threads=4
innodb_write_io_threads=8  #To stress the double write buffer
innodb_buffer_pool_size=2G
innodb_buffer_pool_load_at_startup=ON
innodb_log_file_size = 32M #Small log files, more page flush
innodb_log_files_in_group=2
innodb_file_per_table=1
innodb_log_buffer_size=8M
innodb_flush_method=O_DIRECT
innodb_flush_log_at_trx_commit=0
skip-innodb_doublewrite

Give it a try! I had frame loss issue with only 2 cams and very low load!

I am using 5 ipcam 1280x720 at 25fps and one 1280x960 at 20fps, all in mocord mode and I get no frame loss. I set up event length at 10min, so I have total 14985-15000 frames for each 720p camera at 25fps, and 11980-12000 frames for the 960p @20fps.
Server is Xeon e3-1246V3, and ZM runs into a Esxi 6.5 virtual machine. Virtual machine load does not go over 1.6-2.0!
Hope it can solve part of your problem.

[jzigzone]

marcmerlin:
I know this is an old thread but it might help someone who comes across it. I had the same problem as marcmerlin. Can't see the individual bulk frames when backing up from an Alarm frame. To make this work, first of all you need to select "Frames + Analysis images" in the "Storage" tab of the monitor "Source".

When you view an Alarm frame, the movement is highlighted in red dots. Clicking anywhere in the frame toggles between showing the red dots and not showing the red dots. If you click to not showing the red dots, you can now back up into the previous bulk frames and see them individually!

What you are actually toggling between is the "All Frames" collection of jpgs and the "Analysis (only)" or "Alarmed" set of jpgs. These two collections are stored separately. When in the "Analysis only" section, there are no frames before the "Alarmed" frames as only "Alarmed" frames are in the directory. So when you turn off the red dots, you have actually switched to the "All Frames" directory and you can then proceed backward or forward.