Be Careful with CodeProject AI 2.1.0

[kidrcth]

When I downloaded the new docker image it immediately made my blueiris host shutdown unexpectedly. This corrupted the database. I also noticed I wasn't getting processed image results. So I reverted back the old version (2.0.8) and the results were coming back but a lot of videos from earlier in the day went missing. I did a database restore which took about 4 hours, and i decided to try the update again thinking my database might have been the issue. Nope, reinstalled, my blueiris server had shutdown, came back up and had been missing the videos again. I'm not sure what the issue is, but i wanted to post before someone else installs because there is an update.

Unfortunately you likely wont get much help from me until someone talks about the problem and figures something out. 4 hours to repair the DB is a horrendous price to pay for bug testing.

System specs:
I9-9900k, 32GB RAM
GTX1650 Super
14 Cameras
Blueiris (5.7.4.2) hosted on unrain VM (16GB of RAM)
Codeproject AI hosted in unraid docker








92.0.8

 

[kidrcth]

Below is the log file, the break is where everything suddenly shut down.

0 4/17/2023 1:05:11.468 PM NECam AI: Alert cancelled [nothing found] 29ms
3 4/17/2023 1:05:13.261 PM EastCam MOTION_A
3 4/17/2023 1:05:16.101 PM NECam EXTERNAL
3 4/17/2023 1:05:16.345 PM SideAlley MOTION_A
0 4/17/2023 1:05:18.373 PM EastCam AI: Alert cancelled [nothing found] 29ms
0 4/17/2023 1:05:21.203 PM NECam AI: Alert cancelled [nothing found] 54ms
0 4/17/2023 1:05:21.488 PM SideAlley AI: Alert cancelled [nothing found] 31ms
0 4/17/2023 1:05:25.060 PM EastCam AI: Alert cancelled [nothing found] 29ms
3 4/17/2023 1:05:29.094 PM EastCam MOTION_A
2 4/17/2023 1:05:34.217 PM EastCam AI: not responding
1 4/17/2023 1:05:34.218 PM EastCam AI: Alert cancelled [AI: not responding] 0ms
3 4/17/2023 1:05:47.045 PM SideAlley MOTION_A
3 4/17/2023 1:05:48.084 PM EastCam MOTION_A
2 4/17/2023 1:05:53.350 PM EastCam AI: not responding
1 4/17/2023 1:05:53.354 PM EastCam AI: Alert cancelled [AI: not responding] 5003ms

0 4/17/2023 1:08:56.445 PM App Intel video hardware detected
2 4/17/2023 1:08:56.459 PM App Restarted after unexpected shutdown (5.7.4.2/1500/240)
0 4/17/2023 1:08:57.471 PM MQTT Connect: OK
0 4/17/2023 1:08:57.965 PM App AI has been started
0 4/17/2023 1:08:59.829 PM AI responding at 10.0.0.**:5001
0 4/17/2023 1:09:02.139 PM GarageInside HW VA not compatible: 89
0 4/17/2023 1:09:02.194 PM WestCam HW VA not compatible: 449
0 4/17/2023 1:09:02.215 PM GarageSW HW VA not compatible: 147
0 4/17/2023 1:09:02.453 PM EastCam HW VA not compatible: 91
0 4/17/2023 1:09:02.480 PM Backyard HW VA not compatible: 132
0 4/17/2023 1:09:02.546 PM SideAlley HW VA not compatible: 410
0 4/17/2023 1:09:04.963 PM App Today's sunrise: 6:59 AM, sunset: 7:53 PM
0 4/17/2023 1:09:04.969 PM Server Started on port 81
10 4/17/2023 1:09:27.904 PM local_console [::]: Login
0 4/17/2023 1:09:27.905 PM App Console: connected 1
10 4/17/2023 1:09:27.965 PM ** [::ffff:10.0.0.**]: Login
10 4/17/2023 1:09:28.902 PM ** [::ffff:10.0.0.**]: Login
2 4/17/2023 1:09:35.186 PM Clips db corruption detected; run repair/regenerate
2 4/17/2023 1:09:35.194 PM NWCam Clip: Disk not ready
2 4/17/2023 1:09:35.194 PM GarageSE Clip: Disk not ready
2 4/17/2023 1:09:35.195 PM SWCam Clip: Disk not ready
2 4/17/2023 1:09:35.198 PM Doorbell Clip: Disk not ready
2 4/17/2023 1:09:35.199 PM SECam Clip: Disk not ready
2 4/17/2023 1:09:35.199 PM GarageInside Clip: Disk not ready
2 4/17/2023 1:09:35.205 PM BabyRoom Clip: Disk not ready
2 4/17/2023 1:09:35.212 PM Alley_Int Clip: Disk not ready
2 4/17/2023 1:09:35.219 PM SideAlley Clip: Disk not ready
2 4/17/2023 1:09:35.229 PM WestCam Clip: Disk not ready
2 4/17/2023 1:09:35.236 PM Backyard Clip: Disk not ready
2 4/17/2023 1:09:35.247 PM NECam Clip: Disk not ready
2 4/17/2023 1:09:35.251 PM GarageSW Clip: Disk not ready
2 4/17/2023 1:09:35.256 PM EastCam Clip: Disk not ready

 

[fenderman]

When I downloaded the new docker image it immediately made my blueiris host shutdown unexpectedly. This corrupted the database. I also noticed I wasn't getting processed image results. So I reverted back the old version
(2.0.8) and the results were coming back but a lot of videos from earlier in the day went missing. I did a database restore which took about 4 hours, and i decided to try the update again thinking my database might have been the issue. Nope, reinstalled, my blueiris server had shutdown, came back up and had been missing the videos again. I'm not sure what the issue is, but i wanted to post before someone else installs because there is an update.

Unfortunately you likely wont get much help from me until someone talks about the problem and figures something out. 4 hours to repair the DB is a horrendous price to pay for bug testing.

System specs:
I9-9900k, 32GB RAM
GTX1650 Super
14 Cameras
Blueiris hosted on unrain VM (16GB of RAM)
Codeproject AI hosted in unraid docker








92.0.8

Something is very wrong with your database/recording setup if it takes 4 hours to rebuild it... Do you have hundreds of thousands of files?

 

[kidrcth]

Something is very wrong with your database/recording setup if it takes 4 hours to rebuild it... Do you have hundreds of thousands of files?

I did something a while ago that did this, I went ahead and clicked record on "continuous and triggered" because I was having issues missing some events and hated that all I had left was low quality video ("continuous and alerts). Welllll.... that eats up some space and makes the DB get crazy large, I'm waiting now for that to age out. I went from having like 60 days of video to 3 days.

 

[fenderman]

I did something a while ago that did this, I went ahead and clicked record on "continuous and triggered" because I was having issues missing some events and hated that all I had left was low quality video ("continuous and alerts). Welllll.... that eats up some space and makes the DB get crazy large, I'm waiting now for that to age out. I went from having like 60 days of video to 3 days.

The days of video is irrelevant. Its whether you are using combine and cut properly. You should have one file for every 1-4 hours. Dont cut after each trigger. I can rebuild 8tb of video in a minuet.
Ensure the DB folder is on an ssd.

 

[kidrcth]

The days of video is irrelevant. Its whether you are using combine and cut properly. You should have one file for every 1-4 hours. Dont cut after each trigger. I can rebuild 8tb of video in a minuet.
Ensure the DB folder is on an ssd.

DB is on an SSD, all video is on a HD. Last I checked it said the DB was 5.99GB. Is that normal size?

 

[fenderman]

DB is on an SSD, all video is on a HD. Last I checked it said the DB was 5.99GB. Is that normal size?

No. Mine is less than half a gig for 8TB of recorded video.

 

[kidrcth]

So i have about 7TB of video, and this is my video setting for pretty much all of my cameras.

 

[fenderman]

So i have about 7TB of video, and this is my video setting for pretty much all of my cameras.

View attachment 160236

Delete the entire database folder and let it rebuild itself..
Are you also saving high resolution alert images?

Finally you should probably not be using continuous plus alerts but rather continuous plus triggers.... Otherwise you will not record an high resolution unless there's a confirmed alert but if the AI times out or produces an erroneous result you're going to miss something important.

 

[kidrcth]

Delete the entire database folder and let it rebuild itself..
Are you also saving high resolution alert images?

Finally you should probably not be using continuous plus alerts but rather continuous plus triggers.... Otherwise you will not record an high resolution unless there's a confirmed alert but if the AI times out or produces an erroneous result you're going to miss something important.

I don't think I am saving high resolution images. Continuous plus triggers recorded wayyyy too often. I went from having like 60 days of video to 2-3 days.

 

[fenderman]

I don't think I am saving high resolution images. Continuous plus triggers recorded wayyyy too often. I went from having like 60 days of video to 2-3 days.

You obviously have something set incorrectly

 

[kidrcth]

You obviously have something set incorrectly

What am I misunderstanding?

Continuous plus triggers records anytime a trigger has occurred regardless of if it is AI confirmed. Continuous plus alerts records only when confirmed by AI. The continuous part only applies to the substreams. It records the substream continuously, and when triggered or alerted can record the high res video. With continuous plus triggers a lot more events will be recorded in my experience, exponentially more. Sorry i am a bit confused.

 

[fenderman]

What am I misunderstanding?

Continuous plus triggers records anytime a trigger has occurred regardless of if it is AI confirmed. Continuous plus alerts records only when confirmed by AI. The continuous part only applies to the substreams. It records the substream continuously, and when triggered or alerted can record the high res video. With continuous plus triggers a lot more events will be recorded in my experience, exponentially more. Sorry i am a bit confused.

In your image you have selected continuous plus alerts not continuous plus triggered

 

[kidrcth]

In your image you have selected continuous plus alerts not continuous plus triggered

That's how its set now. I changed it back to that prolly 4-5 days ago. Been waiting for all the crap to just age out.

Database is regenerating, currently on Alert 455500.

 

[fenderman]

That's how its set now. I changed it back to that prolly 4-5 days ago. Been waiting for all the crap to just age out.

Database is regenerating, currently on Alert 455500.

It's set incorrectly now. You want your triggers in high resolution not just confirmed alerts

 

[kjinxx2]

That's how its set now. I changed it back to that prolly 4-5 days ago. Been waiting for all the crap to just age out.

Database is regenerating, currently on Alert 455500.

455,500 Alerts? How big is your Alerts folder set to?

 

[kidrcth]

455,500 Alerts? How big is your Alerts folder set to?

150gb, how much do i want it to be?

 

[kidrcth]

It's set incorrectly now. You want your triggers in high resolution not just confirmed alerts

well alright. ill set it back i guess.

 

[kjinxx2]

150gb, how much do i want it to be?

I have 22TB of storage and my alerts are set to ~10GB - I think this is your issue

@fenderman can probably confirm if that seems problematic

 

[biggen]

I think I have my alerts set to 1GB.