Develop and refine the DVA1622 loader

[spv4u1975]

1 hour ago, majk10 said:

Do you have a problem with enabling push notifications in Ds cam dva1622?

Yes there is a problem as the builds will stop it talking to Synology server by default by editing the hosts I think.

 

I use pushover, instructions can be found here
Tutorial - Synology - Pushover notification integration | SynoForum.com - The Unofficial Synology Forum

 

This give me the notifications and then I setup rules to email me the AI alerts and set it to 5 seconds so you get 3 images of the video to see what is going on.

 

EDIT: actually I couldn't get the text hack to work in SSS yet it works in the main portal, they seem to have upgraded it in SSS, what I di instead was to create a rule for the trigger and then used the pushover API to send a webhook from SSS which then turns into a push https://api.pushover.net/1/messages.xml?user=<yourusername>&token=<your app token>&message=<your+message+to+display+with+pluses+instead+of+spaces+like+this>

Just change the <XX> sections to suit your account and messages

Edited August 2, 2022 by spv4u1975
Error in method

[Menen]

When installed on atom525, the bios is reset. IDE mode is enabled. Further, the error is 58%. There is a solution?

[dimakv2014]

6 минут назад, Menen сказал:

When installed on atom525, the bios is reset. IDE mode is enabled. Further, the error is 58%. There is a solution?

I dont know if it helps, but it works on sata not ide, in fact there is no option IDE at least in my bios to make it work on IDE, since its very old interface and is not supported for a long time

Edited August 3, 2022 by dimakv2014

[marlabs]

Hi Im getting  this error when I try upload photo, any ideas where I can find the logs or what is causing this error ? 

Installation is on proxmox, system has internet connection, downloading packages is working fine. DNS I set to 8.8.8.8 just in case my internal might cause this issue, but without any success. 

 

I was able  to add cameras, but face recognition doesn't work because of this issue. 

 

 

Thanks for advice 

 

 

 

[dimakv2014]

51 минуту назад, marlabs сказал:

Hi Im getting  this error when I try upload photo, any ideas where I can find the logs or what is causing this error ? 

Installation is on proxmox, system has internet connection, downloading packages is working fine. DNS I set to 8.8.8.8 just in case my internal might cause this issue, but without any success. 

 

I was able  to add cameras, but face recognition doesn't work because of this issue. 

 

 

Thanks for advice 

 

 

 

Cpu and igpu info please, if it ls supported then it didn't passthrough

[marlabs]

10 minutes ago, dimakv2014 said:

Cpu and igpu info please, if it ls supported then it didn't passthrough

 

AMD Ryzen 9 3900 MPK + old GeForce 210

 

probably problem is with GPU but the message misleading

[dimakv2014]

6 минут назад, marlabs сказал:

 

AMD Ryzen 9 3900 MPK + old GeForce 210

 

probably problem is with GPU but the message misleading

100% problem with gpu ) it should be intel not nvidia

Edited August 4, 2022 by dimakv2014

[marlabs]

1 minute ago, dimakv2014 said:

100% problem with gpu ) it should be intel not nvidia

this processor doesn't have integrated GPU ... anything else might work ? 

[dimakv2014]

12 минуты назад, marlabs сказал:

this processor doesn't have integrated GPU ... anything else might work ? 

New Dedicated Intel gpus are not tested yet (for example A370) but possible solution, but driver is old and until dsm update or custom driver integration it might not work. So for now the only option is to get Intel cpu with igpu minimum haswell generation

[marlabs]

3 minutes ago, dimakv2014 said:

New Dedicated Intel gpus are not tested yet (for example A370) but possible solution, but driver is old and until dsm update or custom driver integration it might not work. So for now the only option is to get Intel cpu with igpu minimum haswell generation

got it, thank you very much for your time 

[dimakv2014]

22 минуты назад, marlabs сказал:

got it, thank you very much for your time 

Also what I would recommend is to try dva3221 with more recent Nvidia gpu than what you have, for example GTX1650 with gddr5 memory (since gddr6 has different gpu revision which is much newer than the gpu driver that in dva3221), if you can manage install dva3221 it on your amd then with nvidia card gtx1650 you will have more power. I think in your case it will be cheaper and easier just to change gpu than almost whole system.

 

P.s. Alternatively you can just buy cheap MiniPC with Celeron J4125 and use it as baremetal on dva1622, just make sure it has sata and supported ethernet adapter, to check supported lan adapters look in pocopico extensions: https://github.com/pocopico/rp-ext

Edited August 4, 2022 by dimakv2014

[Menen]

I ran two virtual machines on one PC. AI works in both. i 5 7400 + 1650

Edited August 5, 2022 by Menen

[marlabs]

On 8/4/2022 at 7:42 PM, dimakv2014 said:

Also what I would recommend is to try dva3221 with more recent Nvidia gpu than what you have, for example GTX1650 with gddr5 memory (since gddr6 has different gpu revision which is much newer than the gpu driver that in dva3221), if you can manage install dva3221 it on your amd then with nvidia card gtx1650 you will have more power. I think in your case it will be cheaper and easier just to change gpu than almost whole system.

 

P.s. Alternatively you can just buy cheap MiniPC with Celeron J4125 and use it as baremetal on dva1622, just make sure it has sata and supported ethernet adapter, to check supported lan adapters look in pocopico extensions: https://github.com/pocopico/rp-ext

 

I like idea with dva3221, I even built and started installation but I’m getting Something went wrong error with message  We've detected errors on your hard drives (7,8)but I think i saw solution for that

[dimammx]

Hello guys , does the loader support hot plug of HDD. I tried plugging in a disk on hot, but it did not show up. I have bios set up to AHCI, there is NO option for hot plug. Not sure if it supports hot plug, will have to try in windows (need to install it first).   

Any way to check it in DSM, or TC?

Edited August 5, 2022 by dimammx

[dimammx]

Hmm but i was able to hot swap it to a different drive.... Hmm not sure it was just drive issue (the one i installed first failed SMART test and actually disappeared from disks  after an hour) Once i plugged in another one it just showed up. 

[sandro_66]

Hello boys! I want to share great news, a kind of life hack that I received by accident. I have 2 systems: 1) J4125 with 8 GB ram and 2) J4205 with 6 GB ram. The systems are interconnected via CMS. A total of 16 cameras can be added (8 cameras per system). I'm using 12 cameras so far. Both systems have dva1622 builds installed, which work fine, but unfortunately they have a limitation: a maximum of 2 Deep Analytics AI tasks or 1 face recognition task per system. Ok, 4 Deep Analytics AI tasks worked for me accordingly (2 tasks for each system). Surveillance station does not allow me to add more tasks (if already added tasks are enabled). Ok, I wanted to get around this in the following way, in the "Action Rules" create scripts that enable certain Deep AI tasks on certain cameras only when motion detection is triggered on these cameras, for example: motion detection is triggered on the camera in front, the action rule script disables all the "extra" deep ai tasks and turn them on only on the camera in front, and so similarly make scripts for all cameras so that unnecessary deep ai are turned off and only the necessary ones are turned on. But the problem turned out to be that only 5 actions can be set in the action rules and, accordingly, if I prepare deep ai tasks for all 12 cameras in advance, then in these 5 actions I will not be able to set deep ai only on this camera and disable it on the remaining 11 cameras, since I need 12 actions for this, and it is possible to set only 5. In the end, I spat on everything and just installed the script for enabling 2 deep ai tasks on each server and 2 face recognition tasks on the J4125 server when motion detection is triggered on any camera. That is, on the J4125 server, 2 deep ai tasks and 2 face recognition tasks would work in this way (taking into account the fact that in Synology 1 face recognition task = 2 deep ai tasks, it turns out as if 6 deep ai tasks were working, which in principle not allowed (limited) by the dva1623 system). I just decided to try what the system would answer me and expected that it would not execute this scenario and I would see an error in the logs that it was impossible to include so many tasks. However, I was then very surprised that the script of the action rule worked and included all these tasks !!! It turns out this software limitation in Surveillance station. It defines the dva1622 model and artificially limits the ability to MANUALLY enable AI tasks to 2, but there is a loophole in the rules of action so that a script can bypass this by enabling other pre-created AI tasks.

 

Total: now I have 2 face recognition tasks on one server, 2 Deep AI tasks, on the other server so far 2 Deep AI tasks (I haven’t organized scripts in it yet). At the same time, recognition works, tasks are running. I also have many packages running on J4125 (synology drive, synology photo (with face recognition album), application backup, permanent file indexing, and so on). The system does not hang, it works responsively, though the CPU usage shows 99%, but before that it also showed 98-99% (as I understand it, this happens due to file indexing)). Good luck to all!

[dimammx]

1 hour ago, sandro_66 said:

Hello boys! I want to share great news, a kind of life hack that I received by accident. I have 2 systems: 1) J4125 with 8 GB ram and 2) J4205 with 6 GB ram. The systems are interconnected via CMS. A total of 16 cameras can be added (8 cameras per system). I'm using 12 cameras so far. Both systems have dva1622 builds installed, which work fine, but unfortunately they have a limitation: a maximum of 2 Deep Analytics AI tasks or 1 face recognition task per system. Ok, 4 Deep Analytics AI tasks worked for me accordingly (2 tasks for each system). Surveillance station does not allow me to add more tasks (if already added tasks are enabled). Ok, I wanted to get around this in the following way, in the "Action Rules" create scripts that enable certain Deep AI tasks on certain cameras only when motion detection is triggered on these cameras, for example: motion detection is triggered on the camera in front, the action rule script disables all the "extra" deep ai tasks and turn them on only on the camera in front, and so similarly make scripts for all cameras so that unnecessary deep ai are turned off and only the necessary ones are turned on. But the problem turned out to be that only 5 actions can be set in the action rules and, accordingly, if I prepare deep ai tasks for all 12 cameras in advance, then in these 5 actions I will not be able to set deep ai only on this camera and disable it on the remaining 11 cameras, since I need 12 actions for this, and it is possible to set only 5. In the end, I spat on everything and just installed the script for enabling 2 deep ai tasks on each server and 2 face recognition tasks on the J4125 server when motion detection is triggered on any camera. That is, on the J4125 server, 2 deep ai tasks and 2 face recognition tasks would work in this way (taking into account the fact that in Synology 1 face recognition task = 2 deep ai tasks, it turns out as if 6 deep ai tasks were working, which in principle not allowed (limited) by the dva1623 system). I just decided to try what the system would answer me and expected that it would not execute this scenario and I would see an error in the logs that it was impossible to include so many tasks. However, I was then very surprised that the script of the action rule worked and included all these tasks !!! It turns out this software limitation in Surveillance station. It defines the dva1622 model and artificially limits the ability to MANUALLY enable AI tasks to 2, but there is a loophole in the rules of action so that a script can bypass this by enabling other pre-created AI tasks.

 

Total: now I have 2 face recognition tasks on one server, 2 Deep AI tasks, on the other server so far 2 Deep AI tasks (I haven’t organized scripts in it yet). At the same time, recognition works, tasks are running. I also have many packages running on J4125 (synology drive, synology photo (with face recognition album), application backup, permanent file indexing, and so on). The system does not hang, it works responsively, though the CPU usage shows 99%, but before that it also showed 98-99% (as I understand it, this happens due to file indexing)). Good luck to all!

Yea 2 action rules did sound a bit arbitrary. I was kind of expecting synology to   bump this limit in next updates. Perhaps gurus in SS forum can come up with a hack to increase the number unofficially :)

[dimakv2014]

8 часов назад, sandro_66 сказал:

Hello boys! I want to share great news, a kind of life hack that I received by accident. I have 2 systems: 1) J4125 with 8 GB ram and 2) J4205 with 6 GB ram. The systems are interconnected via CMS. A total of 16 cameras can be added (8 cameras per system). I'm using 12 cameras so far. Both systems have dva1622 builds installed, which work fine, but unfortunately they have a limitation: a maximum of 2 Deep Analytics AI tasks or 1 face recognition task per system. Ok, 4 Deep Analytics AI tasks worked for me accordingly (2 tasks for each system). Surveillance station does not allow me to add more tasks (if already added tasks are enabled). Ok, I wanted to get around this in the following way, in the "Action Rules" create scripts that enable certain Deep AI tasks on certain cameras only when motion detection is triggered on these cameras, for example: motion detection is triggered on the camera in front, the action rule script disables all the "extra" deep ai tasks and turn them on only on the camera in front, and so similarly make scripts for all cameras so that unnecessary deep ai are turned off and only the necessary ones are turned on. But the problem turned out to be that only 5 actions can be set in the action rules and, accordingly, if I prepare deep ai tasks for all 12 cameras in advance, then in these 5 actions I will not be able to set deep ai only on this camera and disable it on the remaining 11 cameras, since I need 12 actions for this, and it is possible to set only 5. In the end, I spat on everything and just installed the script for enabling 2 deep ai tasks on each server and 2 face recognition tasks on the J4125 server when motion detection is triggered on any camera. That is, on the J4125 server, 2 deep ai tasks and 2 face recognition tasks would work in this way (taking into account the fact that in Synology 1 face recognition task = 2 deep ai tasks, it turns out as if 6 deep ai tasks were working, which in principle not allowed (limited) by the dva1623 system). I just decided to try what the system would answer me and expected that it would not execute this scenario and I would see an error in the logs that it was impossible to include so many tasks. However, I was then very surprised that the script of the action rule worked and included all these tasks !!! It turns out this software limitation in Surveillance station. It defines the dva1622 model and artificially limits the ability to MANUALLY enable AI tasks to 2, but there is a loophole in the rules of action so that a script can bypass this by enabling other pre-created AI tasks.

 

Total: now I have 2 face recognition tasks on one server, 2 Deep AI tasks, on the other server so far 2 Deep AI tasks (I haven’t organized scripts in it yet). At the same time, recognition works, tasks are running. I also have many packages running on J4125 (synology drive, synology photo (with face recognition album), application backup, permanent file indexing, and so on). The system does not hang, it works responsively, though the CPU usage shows 99%, but before that it also showed 98-99% (as I understand it, this happens due to file indexing)). Good luck to all!

Maybe its possible to create a script to show all recordings in cms from all servers

[sandro_66]

7 часов назад, dimakv2014 сказал:

Maybe its possible to create a script to show all recordings in cms from all servers

Hello!  For some reason, when choosing a server in the recording mode in the "Action Rules" of the AI task or Face Detection in CMS mode, it is reset to the local server: I have the main j4125 server connected to it in the CMS mode, the j4205 server.  If you specify a task from the j4204 server in the actions, when you try to select the j4105 server, it is reset to the local server (that is, the j4125 server).  In fact, in the "Action Rules" I cannot select a camera from the server from the recording mode in CMS (server j4205).

[dimakv2014]

12 минуты назад, sandro_66 сказал:

Hello!  For some reason, when choosing a server in the recording mode in the "Action Rules" of the AI task or Face Detection in CMS mode, it is reset to the local server: I have the main j4125 server connected to it in the CMS mode, the j4205 server.  If you specify a task from the j4204 server in the actions, when you try to select the j4105 server, it is reset to the local server (that is, the j4125 server).  In fact, in the "Action Rules" I cannot select a camera from the server from the recording mode in CMS (server j4205).

Because synology made it this way, I already created feature request, was promised by representative that he will forward it to developers. I am sure its possible to fix this in config files of Surveillance Station using root, at least to choose default server for recordings, but since synology admit they did it this way, its better to go to official forums and make feature request to make it possible see all recording from Local and Recording servers. The more requests, the faster they will fix it. Just dont mention dva or xpenology ))

Edited August 6, 2022 by dimakv2014

[sandro_66]

8 минут назад, dimakv2014 сказал:

Because synology made it this way, I already created feature request, was promised by representative that he will forward it to developers. 

This is great! But no matter how they (Synology) close the loophole with the "Action Rule". I have turned off the automatic updating of packages, but, as I believe, many of them are turned on. Can be turned off unfortunately.

[dimakv2014]

6 минут назад, sandro_66 сказал:

This is great! But no matter how they (Synology) close the loophole with the "Action Rule". I have turned off the automatic updating of packages, but, as I believe, many of them are turned on. Can be turned off unfortunately.

I mean that cms works this way (100% comfirmed by synology), thats why you cant do this in action rule, because default server that shows recordings is Local Host, if they will fix it, then action rules will work as you need.

Edited August 6, 2022 by dimakv2014

[sandro_66]

1 час назад, dimakv2014 сказал:

I mean that cms works this way (100% comfirmed by synology), thats why you cant do this in action rule, because default server that shows recordings is Local Host, if they will fix it, then action rules will work as you need.

yes, maybe my friend. Again, I remember, the main thing would not have covered this gap ...

[dimakv2014]

14 часа назад, sandro_66 сказал:

yes, maybe my friend. Again, I remember, the main thing would not have covered this gap ...

But still, you have done a great research on a problem, I am interested in solution as well one way or another, keep up on a progress ) thanks.

[sandro_66]

В 06.08.2022 в 07:40, sandro_66 сказал:

Hello boys! I want to share great news, a kind of life hack that I received by accident. I have 2 systems: 1) J4125 with 8 GB ram and 2) J4205 with 6 GB ram. The systems are interconnected via CMS. A total of 16 cameras can be added (8 cameras per system). I'm using 12 cameras so far. Both systems have dva1622 builds installed, which work fine, but unfortunately they have a limitation: a maximum of 2 Deep Analytics AI tasks or 1 face recognition task per system. Ok, 4 Deep Analytics AI tasks worked for me accordingly (2 tasks for each system). Surveillance station does not allow me to add more tasks (if already added tasks are enabled). Ok, I wanted to get around this in the following way, in the "Action Rules" create scripts that enable certain Deep AI tasks on certain cameras only when motion detection is triggered on these cameras, for example: motion detection is triggered on the camera in front, the action rule script disables all the "extra" deep ai tasks and turn them on only on the camera in front, and so similarly make scripts for all cameras so that unnecessary deep ai are turned off and only the necessary ones are turned on. But the problem turned out to be that only 5 actions can be set in the action rules and, accordingly, if I prepare deep ai tasks for all 12 cameras in advance, then in these 5 actions I will not be able to set deep ai only on this camera and disable it on the remaining 11 cameras, since I need 12 actions for this, and it is possible to set only 5. In the end, I spat on everything and just installed the script for enabling 2 deep ai tasks on each server and 2 face recognition tasks on the J4125 server when motion detection is triggered on any camera. That is, on the J4125 server, 2 deep ai tasks and 2 face recognition tasks would work in this way (taking into account the fact that in Synology 1 face recognition task = 2 deep ai tasks, it turns out as if 6 deep ai tasks were working, which in principle not allowed (limited) by the dva1623 system). I just decided to try what the system would answer me and expected that it would not execute this scenario and I would see an error in the logs that it was impossible to include so many tasks. However, I was then very surprised that the script of the action rule worked and included all these tasks !!! It turns out this software limitation in Surveillance station. It defines the dva1622 model and artificially limits the ability to MANUALLY enable AI tasks to 2, but there is a loophole in the rules of action so that a script can bypass this by enabling other pre-created AI tasks.

 

Total: now I have 2 face recognition tasks on one server, 2 Deep AI tasks, on the other server so far 2 Deep AI tasks (I haven’t organized scripts in it yet). At the same time, recognition works, tasks are running. I also have many packages running on J4125 (synology drive, synology photo (with face recognition album), application backup, permanent file indexing, and so on). The system does not hang, it works responsively, though the CPU usage shows 99%, but before that it also showed 98-99% (as I understand it, this happens due to file indexing)). Good luck to all!

At the moment, I have 5 AI tasks and 1 face recognition task launched in this way on the DVA1622 assembly based on the J4125 processor (I can set more, but I don’t need this).  All of these tasks run at the same time.  The definition of people, vehicles and face recognition is great.  Now no DVA3221 assemblies and expensive Nvidia GTX1660 video cards are needed for them (to increase the number of AI tasks, since by default only two tasks can be run simultaneously on DVA1622).  Everything works fine based on the J4125 processor and the integrated intel uhd 600 graphics. Synology artificially limited the dva1622 to only use 2 AI tasks at a time.