DVA3221 loader development thread

[spv4u1975]

1 hour ago, RedwinX said:

I can confirm that AI analytics works with a P2000... Didn't have any other graphic card to test, but P2000 works flawlessly...

Interesting, so it is not architecture specific (Pascal Turing) so if enough users report their results we can get an idea of what the common factors/functionality are.

[PaulEvo]

As mentioned in my previous post in this thread the deep video analysis (dva) with a GTX1050 card is not working. One of the dva binaries is restarting in the system and crashes again and so on. After a lot of experiments with a GTX1050 card I have it working in a DVA3221 (7.01 42218 update 3 SS8 and/or SS9 beta in Proxmox) 😅

 

If somenone is interested in how it's done than let me know. Perhaps other GTX10xx series will work as well.

 

Regards

[Orphée]

As mentioned in my previous post in this thread the deep video analysis (dva) with a GTX1050 card is not working. One of the dva binaries is restarting in the system and crashes again and so on. After a lot of experiments with a GTX1050 card I have it working in a DVA3221 (7.01 42218 update 3 SS8 and/or SS9 beta in Proxmox) [emoji28]
 
If somenone is interested in how it's done than let me know. Perhaps other GTX10xx series will work as well.
 
Regards

Just share it here and don't wait for ask.

[PaulEvo]

Hello all,

 

Here we go with the steps voor having DVA (AI) in a DVA3221 with a lower performance video card like a GTX1050

 

1) I assume you have a well functioning SS in DVA3221 7.0.1
2? Download SurveillanceStation-x86_64-8.2.9-6459_DVA_3219.spk   (the 2019 version of DVA, this file is unencrypted)
3) Extract package.tgz from it
4) Extract all the files from the package file which you got in step 3
5) Find the folder and files in it \package\dvacore\etc\model
6) Stop the SS service on your DVA3221
7) Connect with your DVA3221 with WinSCP and go to /volume1/@appstore/SurveillanceStation/dvacore/etc/model

8.) Rename the directory /model to something else
9) Copy the folder and subs which you found in step 5 to it
10) Change ownership and group (recursive) to the 'model' folder and it's contents to 'SurveillanceStation'
11) Start the SS service
12) Create a task in SS in Deep Video Analytics
13) Do some experiments with it and enjoy your tests with AI and the amazing work in all the 'loader' activities

 

Regards, Paul

 

ps you will see a difference in the amount of models in the model directory if you have SS9 beta. More tests are necessary so the model can be identified which is responsible for the incompatibilities before this patch. So for instance the 'PlateDetection' can be used as well with a GTX1050. Will test that in short.

Edited March 23, 2022 by PaulEvo

[jahsoul]

On 3/22/2022 at 8:58 AM, RedwinX said:

I can confirm that AI analytics works with a P2000... Didn't have any other graphic card to test, but P2000 works flawlessly...

 

14 hours ago, PaulEvo said:

As mentioned in my previous post in this thread the deep video analysis (dva) with a GTX1050 card is not working. One of the dva binaries is restarting in the system and crashes again and so on. After a lot of experiments with a GTX1050 card I have it working in a DVA3221 (7.01 42218 update 3 SS8 and/or SS9 beta in Proxmox) 😅

 

If somenone is interested in how it's done than let me know. Perhaps other GTX10xx series will work as well.

 

Regards

 

This is great news. It is good to know that we are not limited to a single GPU and this may open up the door for even more power with SS

[Carl10]

@Orphée@spv4u1975

 

Could anyone who has completed a bare metal install let me know what mods you did for your install. In a previous post I read that pocopico had update the build for the DVA platform, so I tried a standard build and got parse errors. I tried again making the mods I thought were needed and get different parse errors.

 

I am trying to test this build (currently running DS3622XS+) and am having trouble with getting build for a bare metal install. 

This is what I have done:

 

I burned the latest image (tinycore-redpill.v0.4.6.img ), 

Used Putty to continue the customization 

sudo ./rploader.sh update now
sudo ./rploader.sh serialgen DVA3221 now
sudo ./rploader.sh satamap now
sudo ./rploader.sh identifyusb now
sudo ./rploader.sh build denverton-7.0.1-42218

tc@box:~$ sudo ./rploader.sh build denverton-7.0.1-42218
parse error: Expected separator between values at line 319, column 28
parse error: Expected separator between values at line 319, column 28
Error : Platform not found
parse error: Expected separator between values at line 319, column 28
rploader.sh

Version : 0.5.0.1
----------------------------------------------------------------------------------------

 

When I look at the json files with winscp, none of them are updated so started over and modified the custom_config.json to add the denverton build info as outlined by Orphee at the beginning of this thread.

 

and I ran this add and still get a parse error

 

./rploader.sh ext denverton-7.0.1-42218 add https://raw.githubusercontent.com/OrpheeGT/redpill-ext/master/acpid/rpext-index.json
sudo ./rploader.sh

 

and this is what I get:

 

tc@box:~$ ./rploader.sh ext denverton-7.0.1-42218 add https://raw.githubusercont
> ent.com/OrpheeGT/redpill-ext/master/acpid/rpext-index.json
> parse error: Objects must consist of key:value pairs at line 318, column 14
> parse error: Objects must consist of key:value pairs at line 318, column 14
> Error : Platform not found
> parse error: Objects must consist of key:value pairs at line 318, column 14
> rploader.sh
>
> Version : 0.5.0.1
> --------------------------------------------------------------------------------
  --------
> Usage: ./rploader.sh <action> <platform version> <static or compile module> [ext
  ension manager arguments]
>

 

 

If anyone could point me in the right direction I would appreciate the help.

[Orphée]

You must have an issue with a configuration file, maybe user_config.json / custom_config.json / global_config.json...

 

Maybe @pocopico has an idea. You should not have these Parse error.

[Carl10]

@pocopico I believe I have found the problem with the parse errors.  When I first connect to my image on the usb through winscp the custom_config.json file is out of date.  Although the image file has a March 9 file date it does not contain the denverton build information added in the file on your repository.  If I replace that file I no longer get the parse error and can complete the build.  I am now getting disk errors migrating my current 3622 test build even with the drives set to Hot Swappable.  But that is another issue.

 

Conclusion:  the current image does not contain the updated custom_config.json.  The global_config is missing the denverton build as well but not sure what the file does as it worked only updating custom_config.json and is a much older file on your repository. 

[jahsoul]

Quick question; we know that the GPU is there to handle the AI but does our choice of CPU matter for anything? 

[calimansi]

I'm so close on this.  I'm building this in a proxmox environment.

 

I can build it and get it to boot.  I can't get the virtio extension added.  I'm assuming that you need to run:

 

./rploader.sh ext denverton-7.0.1-42218 add https://raw.githubusercontent.com/OrpheeGT/redpill-ext/master/acpid/rpext-index.json

./rploader.sh ext denverton-7.0.1-42218 add https://raw.githubusercontent.com/OrpheeGT/redpill-ext/master/virtio/rpext-index.json

 

before you can run:

 

sudo ./rploader.sh build denverton-7.0.1-42218

 

The problem is if I try to add the extensions before building, I get an error.  If I run:

 

sudo ./rploader.sh clean denverton-7.0.1-42218

 

then it builds, but the extensions don't get added.

 

I'm know I'm doing something the stupid way.  What am I doing wrong.

 

Thanks.

 

[Orphée]

I'm so close on this.  I'm building this in a proxmox environment.
 
I can build it and get it to boot.  I can't get the virtio extension added.  I'm assuming that you need to run:
 
./rploader.sh ext denverton-7.0.1-42218 add https://raw.githubusercontent.com/OrpheeGT/redpill-ext/master/acpid/rpext-index.json
./rploader.sh ext denverton-7.0.1-42218 add https://raw.githubusercontent.com/OrpheeGT/redpill-ext/master/virtio/rpext-index.json
 
before you can run:
 
sudo ./rploader.sh build denverton-7.0.1-42218
 
The problem is if I try to add the extensions before building, I get an error.  If I run:
 
sudo ./rploader.sh clean denverton-7.0.1-42218
 
then it builds, but the extensions don't get added.
 
I'm know I'm doing something the stupid way.  What am I doing wrong.
 
Thanks.
 

Read my posts again and open your eyes too.

[haydibe]

On 3/25/2022 at 8:23 PM, Carl10 said:

tc@box:~$ sudo ./rploader.sh build denverton-7.0.1-42218
parse error: Expected separator between values at line 319, column 28
parse error: Expected separator between values at line 319, column 28
Error : Platform not found
parse error: Expected separator between values at line 319, column 28

Something must be off with your user_config.json. Use "jq '.' ~/user_config.json" (of course without quotes) to see wether your json document is valid. Also you might want to generaly check if the structure is still correct (lookout for missing or missplaced comma or curly bracket character(s)?) and the values look similar to those of other platfrom versions. 

 

Another option is to just attach your user_config.json to a post. Make sure to redact the sn and mac addresses.

 

Update: I should have read the thread to the end. @pocopico the sanity check is missplaced. It needs to be either within "function selectPlatform()" or after it's called the first time. Though, since it sets a global variable, I assume it's safer to place the sanity check it in the function itself.

Edited March 27, 2022 by haydibe

[calimansi]

17 hours ago, Orphée said:

On 3/26/2022 at 3:56 PM, calimansi said:

I'm so close on this.  I'm building this in a proxmox environment.
 
I can build it and get it to boot.  I can't get the virtio extension added.  I'm assuming that you need to run:
 
./rploader.sh ext denverton-7.0.1-42218 add https://raw.githubusercontent.com/OrpheeGT/redpill-ext/master/acpid/rpext-index.json
./rploader.sh ext denverton-7.0.1-42218 add https://raw.githubusercontent.com/OrpheeGT/redpill-ext/master/virtio/rpext-index.json
 
before you can run:
 
sudo ./rploader.sh build denverton-7.0.1-42218
 
The problem is if I try to add the extensions before building, I get an error.  If I run:
 
sudo ./rploader.sh clean denverton-7.0.1-42218
 
then it builds, but the extensions don't get added.
 
I'm know I'm doing something the stupid way.  What am I doing wrong.
 
Thanks.
 

Read my posts again and open your eyes too.

 

 

Yep... you were right.  Just a hint that the ext add command have different http addresses.  Thanks.  No assumptions.

[Carl10]

On 3/13/2022 at 3:49 PM, Orphée said:

- Surveillance Station is able to run without any GPU and still work for standard camera features like normal NAS, but has 8 licences available instead of 2 (but logs will flood in /var/log/messages and /var/log/kern.log about missing GPU, maybe a risk of heavy log generation and disk space).

@Orphée I have successfully migrated my bare metal 3622 test machine into a DVA3221.  I started with a GTX 750i card to try, and it was recognized.  The AI can be set up but doesn't generate any results.  I have looked at the logs and below is a portion of what I saw.  Are these the errors you saw with no card?

 

Since I migrated, my SS 9 beta was moved to the new setup which may be a different version then the AI packages are expecting.  I am in the process of backing up my current setup, then I will delete the Beta SS and install the standard version to see if that corrects the problem.  Did your install have SS 9 as standard or Beta?

 

I don't know enough about that error to know if it is due to the SS package or the video card library.

 

I will also try @PaulEvotrick with DVA3219 model folder to see if that eliminates these errors.

 

If all else fails I will have to get a  1050 or 1650 card.

 

Thanks in advance for any light you can shed.

 

 

2022-03-27T19:32:27-05:00 DSM7 kernel: [18086.435407] synofaced[8040]: segfault at 0 ip 00007fa7925fba60 sp 00007fa710010618 error 4 in libdvacore-dl.so[7fa7925e5000+57000]
2022-03-27T19:32:27-05:00 DSM7 coredump[8087]: Process synofaced[8026](/volume1/@appstore/SurveillanceStation/synoface/bin/synofaced) dumped core on signal [11]. Core file [/volume1/@synofaced.core.gz]. C
mdline [/var/packages/SurveillanceStation/target/synoface/bin/synofaced ]
2022-03-27T19:32:31-05:00 DSM7 coredump[8077]: Core file [/volume1/@synodvad.core.gz] size [66875288]
2022-03-27T19:32:31-05:00 DSM7 coredump[8087]: Core file [/volume1/@synofaced.core.gz] size [65804743]
2022-03-27T19:32:37-05:00 DSM7 kernel: [18096.434908] synodvad[8329]: segfault at 0 ip 00007f0124386a60 sp 00007f00a8020a08 error 4 in libdvacore-dl.so[7f0124370000+57000]
2022-03-27T19:32:37-05:00 DSM7 coredump[8340]: Process synodvad[8317](/volume1/@appstore/SurveillanceStation/synodva/bin/synodvad) dumped core on signal [11]. Core file [/volume1/@synodvad.core.gz]. Cmdli
ne [/var/packages/SurveillanceStation/target/synodva/bin/synodvad ]
2022-03-27T19:32:37-05:00 DSM7 kernel: [18096.538038] synofaced[8301]: segfault at 0 ip 00007f684cd4aa60 sp 00007f67d0010618 error 4 in libdvacore-dl.so[7f684cd34000+57000]
2022-03-27T19:32:37-05:00 DSM7 coredump[8355]: Process synofaced[8285](/volume1/@appstore/SurveillanceStation/synoface/bin/synofaced) dumped core on signal [11]. Core file [/volume1/@synofaced.core.gz]. C
mdline [/var/packages/SurveillanceStation/target/synoface/bin/synofaced ]
2022-03-27T19:32:41-05:00 DSM7 coredump[8340]: Core file [/volume1/@synodvad.core.gz] size [66250202]
2022-03-27T19:32:41-05:00 DSM7 coredump[8355]: Core file [/volume1/@synofaced.core.gz] size [65469715]
2022-03-27T19:32:47-05:00 DSM7 kernel: [18106.537744] synodvad[8668]: segfault at 0 ip 00007f02874eba60 sp 00007f0210020a08 error 4 in libdvacore-dl.so[7f02874d5000+57000]
202

[calimansi]

14 hours ago, calimansi said:

 

 

Yep... you were right.  Just a hint that the ext add command have different http addresses.  Thanks.  No assumptions.

 

@Orphée - Thank you so much for putting this together.  This works great!

 

I am just doing some testing.  As stated, I am running this in proxmox... no GPU.  I have (3) wyze cams connected to it using the wyze bridge docker in Home Assistant.  This thing is damn stable.  Ran all night recording different events happening around me.  The way it presents the videos is perfect... way better than anything wyze has done.

 

I was going to use frigate with a google coral, but this is so much more polished.

 

I am planning on getting a GPU... probably roll with the GTX1650.  I do have an ancient quadro k420, but that might be a tad too old.  I'm not sure if it even supports CUDA.

 

Thank you for everyone else for doing a lot of testing.

[kylaris]

Using PaulEvo trick for 10 series. I've gotten GTX 1080 ti to work but there are problems with Surveillance Station. Probably due to having Frankenstein app.

For example, the face recognition works and records the faces but when I try to register the face as user, I get an error (forgot the exact error), I think it was like " network is unstable".  

PaulEvo, Are you getting such error with GTX 1050?

Also for people with GTX 1650, everything is function as it should? no errors? 

 

Thank you.

Edited March 29, 2022 by kylaris

[Orphée]

Yes, everything works as it should for me with gtx1650.

[Carl10]

On 3/14/2022 at 6:11 AM, PaulEvo said:

Replacing it with a GTX1650 works flawless.

@PaulEvo@Orphée  When you put your 1650 cards into your setup was that onto a clean test machine?  Or did you add/replace the 1650 card into a DVA build that had no card or a different card?

 

I am asking because I could not get my 750ti to work (which was recognized but got the same error Paul saw) and replaced it with a 1650 card in the same build and now in the info center the GPU shows speed and memory but list the card as [unkown error].

 

I tried to rebuild the boot drive and install on the current system but it just connected right back into my old build with the same error.

 

Is there a way to reset the card information (erase the 750ti)

 

Thanks for any help.

[kylaris]

8 hours ago, kylaris said:

Using PaulEvo trick for 10 series. I've gotten GTX 1080 ti to work but there are problems with Surveillance Station. Probably due to having Frankenstein app.

For example, the face recognition works and records the faces but when I try to register the face as user, I get an error (forgot the exact error), I think it was like " network is unstable".  

PaulEvo, Are you getting such error with GTX 1050?

Also for people with GTX 1650, everything is function as it should? no errors? 

 

Thank you.

I think I fixed the unstable network error for face recognition by replacing the /synoface/etc from DVA3219 

[merve04]

Does DVA3221 loader play nice with plex for hardware transcoding? Also I don’t have a dedicated GPU, just whatever is in a i5-8400. I’m just interested in the fact SS will allow 8 cameras. 

[pocopico]

For anyone with a test DVA3221, i have tested succesfully the TCRP two stage upgrade to 7.1.0-42621.

 

Please proceed with caution and only on test systems.

 

The process is :

- Manual upgrade using DSM to 7.1 RC
- On reboot, select to boot tinycore redpill
- While on TCRP run the following

 

./rploader.sh update now
./rploader.sh backuploader now
./rploader.sh download 
./rploader.sh download denverton-7.0.1-42218
./rploader.sh postupdate denverton-7.0.1-42218

 
Fix the acpi systemd issue otherwise DSM will shutdown on access

 

./rploader.sh mountdsmroot now

sed -i 's/^acpi-cpufreq/# acpi-cpufreq/g' /mnt/dsmroot/usr/lib/modules-load.d/70-cpufreq-kernel.conf && cat /mnt/dsmroot/usr/lib/modules-load.d/70-cpufreq-kernel.conf

 

for people NOT having an NVIDIA card the following sould also be performed otherwise DSM will shutdown on access

 

Remove both lines on :  /mnt/dsmroot/usr/lib/modules-load.d/70-syno-nvidia-gpu.conf

 

[Carl10]

 

Could anyone help with an "unknown error" where the GPU is listed in the DVA3221? I initially did a bare metal DVA3221 install and tried a GTX 750ti card (it is expecting GTX 1650).  It recognized the card but generated repeated errors and could not perform any of the deep AI.  I swapped out the 750ti for a brand new 1650 card and the card is not recognized giving me "unknown error" instead of the card model. It did show the correct GPU speed,  memory and temperature information.  I have rebuilt the machine with a fresh drive and get the same result:

 

 

 

The only NVIDIA reference in the kernel and message logs is shown below.  I did not add any extensions during the build process.

 

2022-03-30T20:13:37-05:00 DSM7 kernel: [   38.080924] 
2022-03-30T20:13:37-05:00 DSM7 kernel: [   38.162079] lp3943 0-0060: failed to read 0x00
2022-03-30T20:13:37-05:00 DSM7 kernel: [   38.162080] lp3943 0-0060: chip detection err: -6
2022-03-30T20:13:38-05:00 DSM7 umount[7332]: can't umount /initrd: Invalid argument
2022-03-30T20:13:38-05:00 DSM7 kernel: [   43.291775] NVRM: Persistence mode is deprecated and will be removed in a future release. Please use nvidia-persistenced instead.
2022-03-30T20:13:38-05:00 DSM7 synostorage[7744]: synostorage_int_disk.c:434 Disk [sda] info cache generated

 

 

 

2022-03-30T20:13:28-05:00 DSM7 kernel: [   29.865557] EXT4-fs (md0): couldn't mount as ext3 due to feature incompatibilities
2022-03-30T20:13:28-05:00 DSM7 kernel: [   33.712441] flashcache: flashcache-1.0-2021-10-18-20:42 initialized
2022-03-30T20:13:28-05:00 DSM7 kernel: [   33.721233] flashcache_syno: flashcache-1.0-syno-v25-pin-file-2021-10-18-20:42-debug initialized
2022-03-30T20:13:29-05:00 DSM7 syno-coredump-store.sh[7030]: Set coredump path to [/var/crash]
2022-03-30T20:13:29-05:00 DSM7 syno_swap_ctl[7042]: swap/swap_state_check.c:23 swap/swap_state_check.c(23): Swap state(None) is not Ready!
2022-03-30T20:13:29-05:00 DSM7 synocheckuser[7074]: synocheckuser.c:1043 start to check user data
2022-03-30T20:13:29-05:00 DSM7 synocheckuser[7074]: synocheckuser.c:1116 check user data finished
2022-03-30T20:13:31-05:00 DSM7 kernel: [   36.956256] NVRM: loading NVIDIA UNIX x86_64 Kernel Module  440.44  Sun Dec  8 03:38:56 UTC 2019
2022-03-30T20:13:32-05:00 DSM7 synonetseqadj[7190]: synonetseqadj.c:421 Error internal NIC devices 1 does not equal to internal NIC number 4
2022-03-30T20:13:37-05:00 DSM7 kernel: [   38.069386] Get empty minor:104
2022-03-30T20:13:37-05:00 DSM7 kernel: [   38.078377] 80 29 12 a0 ff ff ff ff  [00] 0x000     ffffffffa0122980    __this_module+0x0/0xffffffffffffb171 [denverton_synobios]
2022-03-30T20:13:37-05:00 DSM7 kernel: [   38.078380] 60 66 11 a0 ff ff ff ff  [01] 0x008     ffffffffa0116660    GetBrand+0x0/0x10 [denverton_synobios]
2022-03-30T20:13:37-05:00 DSM7 kernel: [   38.078384] 90 c8 11 a0 ff ff ff ff  [02] 0x010     ffffffffa011c890    GetModel+0x0/0x1e0 [denverton_synobios]
2022-03-30T20:13:37-05:00 DSM7 kernel: [   38.078388] 00 00 00 00 00 00 00 00  [03] 0x018               (null)              (null)
 

There are two versions of this card one with DDR5 memory and mine with DDR6.

 

Anyone have any ideas what I missed or where to look for a solution.

 

Thanks for any help.

 

[ultra]

Great jobs.I want to make a loader ds716+,can u tell me how to make a new loader?Thank u very much

[spv4u1975]

On 3/29/2022 at 6:47 PM, Carl10 said:

When you put your 1650 cards into your setup was that onto a clean test machine?  Or did you add/replace the 1650 card into a DVA build that had no card or a different card?

I had a 1030 card on mine at build but it didn't work so I ordered a 1650, when it arrived I just shut down NAS swapped the card and rebooted. New model was there in reboot and AI started working immediately

 

I didn't attempt any of the fixes advised above as they didn't exist so every was as per original install of pat file.

[spv4u1975]

On 3/30/2022 at 4:10 AM, merve04 said:

I’m just interested in the fact SS will allow 8 cameras.

Yes it does I've got 6 IP cameras going with no additional licences required.

 

Continuous recording of low quality stream and then recording high quality on motion, then AI settings as well to identify what caused the motion.

 

I'm also running the SSS beta v9 fine.

 

As push notifications will not work to DS Cam I setup pushover and an action rule so the SSS rule pushes to the alerts to that service by a webhook to get AI notifications as push on mobile.

Edited March 31, 2022 by spv4u1975