"Signal Lost/reacquiring" wired IP Cams

[nugget]

Hi all,

After an absense of many years I've decided to reinstall Zoneminder at my new house, and have some annoying difficulties with a wired IP Camera (Ebay special - Tenvis IP602W for abt. $AU80).

The picture quality is what I expected for the price, but I'm having some difficulties with zero byte frames coming through regularly (the "blue blip" that has been mentioned in the forums in the past). At the time of the "blip" the log file also reports:

Code: Select all

Mar 13 15:08:16 zoneminder zma_m1[15356]: WAR [Signal: Lost]
Mar 13 15:08:16 zoneminder zma_m1[15356]: INF [Driveway: 14546 - Closing event 127, signal loss]
Mar 13 15:08:17 zoneminder zma_m1[15356]: WAR [Signal: Reacquired]

Several fixes have been suggested (ring buffer size, network bandwidth, memory segments, etc), none of which I think applies to me.

Given my last experience with Zoneminder was 0.9.15, I thought I'd ask here. Failing finding an actual fix for this issue, which I think is probably something to do with the http acquiring of the images from the camera, I'd like to simply ignore any event which is 1 alarm frame long. Is there an easy way to do this?

Worst case, I used to have a script I ran which looked for 1 frame events and deleted them from the database/filesystem, however the location of the image files is now a little more complex than it used to be (/var/www/htmp/zm/eve/t/etc/etc/etc). Scriptable, but annoying.

Has anyone created a cleanup routine to delete these "small events" from the DB/filesystem? This might save me a major rewrite/test to my script.

My configuration is as follows:
ZM Server - Centos 6.5 on VMware ESXi
DB server - Centos 6.5 on VMware ESXi
(the above 2 virtual machines run on a HP Microserver, not the most powerful "server" available but really really really power efficient)
IP camera 100MB wired network connection

ZM monitor is configured to acquire images via http, using the onboard videostream.cgi at a frame rate of 5 FPS

I do not believe I have a performance issue on the ESXi server or guests - there is no evidence of CPU/memory/disk bottlenecks.

[mikb]

I'd like to simply ignore any event which is 1 alarm frame long. Is there an easy way to do this?

If you can't fix the underlying problem of the blue-blips, on the zone-area settings dialogue, for each erratic camera, try "Overload Frame Ignore Count: 2" see if that calms it down?

[nugget]

Some feedback on this one. I have not solved the "blue-blips" but have managed to calm down the number of alarms I'm receiving.

Not only did I add "Overload Frame Ignore Count: 2" to the zone as suggested by mikb, but I also increased "Alarm frame count:2" on the monitor.

Tweaking will continue, I've got the number of events down from hundreds a day to a couple of dozen.

[mikb]

Not only did I add "Overload Frame Ignore Count: 2" to the zone as suggested by mikb, but I also increased "Alarm frame count:2" on the monitor.

Tweaking will continue, I've got the number of events down from hundreds a day to a couple of dozen.

Ack! I forgot that one -- I was looking on the monitor's config page but didn't see it!

Good to see that things have improved ...