Possible Memory leak

[wpiman]

I have approximately 12 cameras in BI with BI tools installed with mostly Duhua cameras. All direct to disk-- using substreams. Backups are to an on network NAS but the direct to write disk is local. Release 5.4.9.16

Everything works great-- for a while. When I started BI it sips maybe 10-20% CPU and around 1.7 Giga of memory. Everything is smooth.

This morning, I was reviewing footage and I noticed the footage was choppy. I noticed the CPU was up around 70-80% and the memory usage was at 88%. This is on a machine with 32 Gigs of RAM. I logged in the the task manager showed it was at 24 Gigs.

Could this be due to backups occurring? I am thinking about backing out to an earlier version....

Thanks,

 

[sebastiantombs]

Check your video driver version. I'm running 20+ cameras on version .16. Memory and CPU utilization are stable. There are numerous threads with information on Intel video drivers, have a look around. It also could be related to your backup, but I doubt it.

 

[wpiman]

Check your video driver version. I'm running 20+ cameras on version .16. Memory and CPU utilization are stable. There are numerous threads with information on Intel video drivers, have a look around. It also could be related to your backup, but I doubt it.

Thanks. I will double check that. I know I had it set correctly at one point but perhaps windows thought it knew better than me. Shutting off all of the auto updates has not been trivial and maybe one slipped by me.

 

[trooperthorn]

If you feel like you have a Memory usage issue, it is always best to observe and maintain a continuous monitoring platform in which you can run reports against. Several companies offer free tools to monitor CPU, Memory, and Volume utilization for small deployments. Keep and monitor the last few months of servers, and services. Checking Utilization for CPU you will see normal spikes at sun up and quiets down at sundown.

Memory Leaks are super easy to spot with Continuous monitoring, it results in a Sawtooth utilization over time. Regular Sawtooth (Memory balloons from start to restart) & Reverse Sawtooth (starts high and slowly lowers over time) are the 2 that you look for.

System OS Hardware is not as important as the decoder used for the cameras. I run continuous recording, and from what I have found, Every 6-10 cameras should be dedicated to a Nvidia P600/P1000 or similar card.

Nvidia P series cards allow you to run WMI queries on the cards to obtain utilization for further continuous monitoring. K series does not do WMI, nor should they be used. Remember that you need to get a HDMI Dummy plug to activate the card when a monitor is not connected.

TLDR; Start Continuous monitoring, get Nvidia P series cards, get HDMI Dummy.

 

[wpiman]

If you feel like you have a Memory usage issue, it is always best to observe and maintain a continuous monitoring platform in which you can run reports against. Several companies offer free tools to monitor CPU, Memory, and Volume utilization for small deployments. Keep and monitor the last few months of servers, and services. Checking Utilization for CPU you will see normal spikes at sun up and quiets down at sundown.

Memory Leaks are super easy to spot with Continuous monitoring, it results in a Sawtooth utilization over time. Regular Sawtooth (Memory balloons from start to restart) & Reverse Sawtooth (starts high and slowly lowers over time) are the 2 that you look for.

System OS Hardware is not as important as the decoder used for the cameras. I run continuous recording, and from what I have found, Every 6-10 cameras should be dedicated to a Nvidia P600/P1000 or similar card.

Nvidia P series cards allow you to run WMI queries on the cards to obtain utilization for further continuous monitoring. K series does not do WMI, nor should they be used. Remember that you need to get a HDMI Dummy plug to activate the card when a monitor is not connected.

TLDR; Start Continuous monitoring, get Nvidia P series cards, get HDMI Dummy.

Thanks.

I have set my software to run the latest stable update and I have been running now for about 3 weeks and the memory usage has stabilized...