There's got to be a fix for this. Deepstack runs for a month or so then starts timing out on every camera causing no alerts. No settings changed whatsoever. Uninstall/reinstall DS and all is well.
DEEPSTACK STILL TIMING OUT, NO ALERTS
- Thread starter flyboy68
- Start date
shannondalebreaux
Getting the hang of it
No I haven't. But when installing i'd think the latest version would be installed? Only runs for a month or so before timing out again
shannondalebreaux
Getting the hang of it
I had the same trouble. I updated DS and it fixed mine not sure if that will help yours.
sebastiantombs
Known around here
sebastiantombs
Known around here
I also run the exact same version of DS, CPU only, and haven't run into this issue (but I do occasionally reboot the server for other reasons, perhaps every two weeks or so). That being said, I believe the DS developers acknowledged an issue with this version related to this, about a month or two ago, and hope to have it resolved in the next release. I know some here seem to prefer the 09/21 dated DS release.
On a related note, this makes me wonder if that is why Ken at BI has decided to move over to SenseAI as the "primary/recommended" AI vision solution, as their development schedule seems very aggressive (somewhat matching his own rapid release pattern). Just a possibility, nothing confirmed.
On a related note, this makes me wonder if that is why Ken at BI has decided to move over to SenseAI as the "primary/recommended" AI vision solution, as their development schedule seems very aggressive (somewhat matching his own rapid release pattern). Just a possibility, nothing confirmed.
sebastiantombs
Known around here
Yes, the 2022.01.01 seems to have problems. Revert back to the 2021.02.01 version and see how that works out.
sebastiantombs
Known around here
Gotta improve your googlefu
github.com
Releases · johnolafenwa/DeepStack
The World's Leading Cross Platform AI Engine for Edge Devices - johnolafenwa/DeepStack
github.com
Futaba
Pulling my weight
- Nov 13, 2015
- 227
- 168
I had this problem earlier this week after I switched over to general v3.1. one thing I did that might have resolved the problem was I added an object name to the cancel box for the stuck camera. That kind of unclogged the pipe. But it could be totally a coincidence.
ipcamtalk.com
Deepstack occupied ignore static object
Hello all, I am trying to configure deepstack inside BI and something weird its happening. I have selected the ignore static object, but it seems that even after having the option to detect person in DS, that is being bypassed somehow: As you can see, it is ignoring the person even at 87%...
DanDenver
Getting comfortable
Funny, I just had a similar issue. For about 16 months my setup was fine. Then, about 3 weeks ago, DS started reporting 'timeout' and not sending alerts.
I have new very robust hardware so it simply cannot be related to that (dedicated PC built specifically for BI and BI alone that was custom built following all the guidelines on this forum, for about $2400.00. I have 15 cameras and CPU hangs around 5%-7%).
Anyway, my BI license had relapsed around this same time and some posts indicated that alerts went away when that expired. So I renewed my license - no change.
Then I versioned up BI to the latest: 5.5.6.21 but no joy.
Then I versioned up CPU DS from: 2021.02.1 to CPU 2022.01.1, and just like, that everything started working again and has so for the last two weeks.
Yes I restarted my machine, BI and Deepstack in parallel as well independently and in several combinations before versioning anything up on any software (we are talking about windows here!).
I reviewed the OS/BI logs for clues. Once I decided it was DS itself I tried to issue any command that would let me see if it was having a memory constrained issue, but my best guess after reading about that was that to perform such commands required me to deploy DS into a container. While I am technical and am a bit twiddler at work, I try to keep that crap at work if at all possible so chose not to pursue that route. In the end, the versioning up of DS seemed to do the trick for me.
Keeping my fingers crossed. Would really prefer to know root cause however. I did not know that BI was pivoting on what AI it recommends/supports.
I have new very robust hardware so it simply cannot be related to that (dedicated PC built specifically for BI and BI alone that was custom built following all the guidelines on this forum, for about $2400.00. I have 15 cameras and CPU hangs around 5%-7%).
Anyway, my BI license had relapsed around this same time and some posts indicated that alerts went away when that expired. So I renewed my license - no change.
Then I versioned up BI to the latest: 5.5.6.21 but no joy.
Then I versioned up CPU DS from: 2021.02.1 to CPU 2022.01.1, and just like, that everything started working again and has so for the last two weeks.
Yes I restarted my machine, BI and Deepstack in parallel as well independently and in several combinations before versioning anything up on any software (we are talking about windows here!).
I reviewed the OS/BI logs for clues. Once I decided it was DS itself I tried to issue any command that would let me see if it was having a memory constrained issue, but my best guess after reading about that was that to perform such commands required me to deploy DS into a container. While I am technical and am a bit twiddler at work, I try to keep that crap at work if at all possible so chose not to pursue that route. In the end, the versioning up of DS seemed to do the trick for me.
Keeping my fingers crossed. Would really prefer to know root cause however. I did not know that BI was pivoting on what AI it recommends/supports.
Running BI 5.5.6.21, DS 2021.02.1 Uninstall DS, delete DS folder, install again. It will work for a while. I can get 3mo+ out of it by doing this.
It saves a ton of garbage in the DS folder.
It saves a ton of garbage in the DS folder.
DanDenver
Getting comfortable
Is there a specific folder that can be cleaned out? Re-installing is so drastic. If a specific folder is determined to be the trouble maker, it would be so much quicker and could then be cleaned out by a script that is on a scheduler.