docker-blueiris
docker-blueiris copied to clipboard
jshridha
Webserver stops responding
Hi,
I've got docker-blueiris and node-deepstackai-trigger set up, inspired by this Hook Up video. I'm at the stage where BlueIris feeds low-res stills to deepstack, which can trigger high-res recordings in BlueIris. It works perfectly... if it wasn't for the BlueIris webserver being incredibly flakey. I'm new to BlueIris and don't know it's history, but it looks like this thread may refer to the same issue I'm having.
Sometimes, after starting docker-blueiris, the web UI works fine, but then it'll stop responding. Sometimes it doesn't work at all after starting the docker container. I've tried turning the webserver off and on again, or changing the port number. Looking at netstat, I've seen BI continue to listen on the original port (even though it's no longer responding), even after I've turned off the webserver or changed its port number. And on other occasions, changing the port number will make the web UI work again (on the new port). Restarting the docker container sometimes fixes the issue, sometimes not. I've also tried running it on the host network, in case it was an issue in docker, but I still saw the issue.
I tried switching to BlueIris version 4, but the install didn't work. I tried twice with the same error.
So, my questions are... is the web UI flakey for everyone else, and does the version 4 install still work for anybody else, or has it been broken but nobody's noticed because nobody uses it?
I'm so close to having all this working :)
Thanks,
Chris
I set up docker-blueiris and node-deepstackai-trigger once more from scratch and it worked great (triggered HD recordings at the correct times) before dying less than 24 hours later. The docker-blueiris container is still running, but it's no longer responding to VNC or the web interface. It looks like the Blue Iris process has died, but the docker container hasn't restarted because the main process is supervisord, which is still running fine. Also, supervisord's config for blueiris contains autorestart=false - is that correct?
Blue Iris log below (I named my 'aiinput' directory 'Stills'). I don't know what happened at 3:30 pm, but AFAIK, the network's been fine - I've been on a video call all day and the camera's fine at the moment. And besides, any network / camera outage shouldn't result in Blue Iris dying.
3 11/17/2020 2:29:12.371 PM patio-sd MOTION
3 11/17/2020 2:29:12.669 PM patio-hd EXTERNAL
10 11/17/2020 2:30:23.807 PM aiuser ::ffff:172.17.0.1: Logout, 2:01
0 11/17/2020 2:33:36.977 PM New Move: nothing to do [19/168 hrs, 1.58G/10.0G, 760.7G free]
0 11/17/2020 2:33:36.980 PM Stills Delete: nothing to do [19/168 hrs, 6.96M/1.00G, 760.7G free]
0 11/17/2020 2:43:37.635 PM New Move: nothing to do [19/168 hrs, 1.59G/10.0G, 760.7G free]
0 11/17/2020 2:43:37.640 PM Alerts Delete: nothing to do [18/168 hrs, 0K/1.00G, 760.7G free]
0 11/17/2020 2:53:38.306 PM New Move: nothing to do [19/168 hrs, 1.60G/10.0G, 760.7G free]
0 11/17/2020 3:03:38.977 PM New Move: nothing to do [20/168 hrs, 1.61G/10.0G, 760.6G free]
0 11/17/2020 3:03:38.981 PM Stored Delete: nothing to do [0/336 hrs, 0K/20.0G, 760.6G free]
0 11/17/2020 3:03:38.984 PM Alerts Delete: nothing to do [18/168 hrs, 0K/1.00G, 760.6G free]
0 11/17/2020 3:03:38.987 PM Stills Delete: nothing to do [19/168 hrs, 6.96M/1.00G, 760.6G free]
0 11/17/2020 3:08:39.332 PM Stills Delete: nothing to do [20/168 hrs, 6.96M/1.00G, 760.6G free]
0 11/17/2020 3:13:39.680 PM New Move: nothing to do [20/168 hrs, 1.62G/10.0G, 760.6G free]
0 11/17/2020 3:23:40.415 PM New Move: nothing to do [20/168 hrs, 1.63G/10.0G, 760.6G free]
1 11/17/2020 3:30:34.552 PM patio-sd Signal: I/O error: 0
1 11/17/2020 3:30:34.709 PM patio-sd Signal: network retry
4 11/17/2020 3:30:34.917 PM patio-sd Signal: restored
1 11/17/2020 3:30:35.042 PM patio-sd Signal: network retry
4 11/17/2020 3:30:35.043 PM patio-sd Signal: restored
1 11/17/2020 3:30:35.213 PM patio-hd Signal: network retry
4 11/17/2020 3:30:35.338 PM patio-hd Signal: restored
1 11/17/2020 3:42:48.053 PM patio-sd Signal: I/O error: 0
1 11/17/2020 3:42:48.045 PM patio-sd Signal: network retry
1 11/17/2020 3:42:49.560 PM patio-sd Signal: network retry
1 11/17/2020 3:42:54.794 PM patio-hd Signal: network retry
4 11/17/2020 3:42:59.006 PM patio-hd Signal: restored
0 11/17/2020 3:43:00.230 PM Alerts Delete: nothing to do [19/168 hrs, 0K/1.00G, 760.6G free]
4 11/17/2020 3:43:10.386 PM patio-sd Signal: restored
0 11/17/2020 3:43:49.839 PM App Console: has died
It appears to be still running in docker (up 25 hours, so it hasn't restarted). blueiris-dev is my build of the head of the dev branch as it stands today (a4013b3). I have the Blue Iris web UI running on port 8086 and VNC exported on 5901.
me@desktop:~/cameras/blueiris/drive_c/BlueIris/log$ docker ps
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
ecc1ac735756 danecreekphotography/node-deepstackai-trigger:latest "node --no-deprecati…" 23 hours ago Up 23 hours 0.0.0.0:4242->4242/tcp node-deepstackai-trigger_trigger_1
1e5c58416cab deepquestai/deepstack:latest "/app/server/server" 23 hours ago Up 23 hours 5000/tcp node-deepstackai-trigger_deepstack-ai_1
a10db056258b blueiris-dev "/usr/bin/supervisor…" 25 hours ago Up 25 hours 0.0.0.0:8080->8080/tcp, 0.0.0.0:8086->8086/tcp, 0.0.0.0:5901->5900/tcp blueiris
Here's the docker log (it's full of get_date_format logs):
0694:fixme:variant:get_date_format ignoring flags 402
0694:fixme:variant:get_date_format ignoring flags 402
01c4:fixme:variant:get_date_format ignoring flags 402
0694:fixme:variant:get_date_format ignoring flags 402
0694:fixme:variant:get_date_format ignoring flags 402
/home/wineuser/blueiris.sh: line 31: 267 Killed wine "${BLUEIRIS_EXE}"
0694:fixme:variant:get_date_format ignoring flags 402
2020-11-17 15:43:49,613 INFO exited: blueiris (exit status 137; not expected)
I included a link to a thread in my original post about other people seeing BlueIris dying. Again, I'm not sure if it's the same issue that's affecting me, but I'm guessing Blue Iris (running on Windows) is pretty reliable for most people. We're running it under Wine in Docker, which adds another couple of levels of things to go wrong, but I'm guessing docker-blueiris is more reliable than I'm finding for most people (although it's clearly we're a much smaller crowd).
I'm running a stock version of Ubuntu 20 with not much else running on the box.
Any clues for how to debug further or other suggestions welcome.
So exit code 137 appears to be related to OOM. The host has 32GB RAM, of which I'm only using 5, so plenty free.
I don't know Docker that well, but could it be that not enough memory has been granted to the container? I'm clutching at straws now, because I only have one camera set up in BI, so I doubt the default allocation is less than what's needed for one camera, but hey...
