Memory leak on version 5.8.0.5

[awitty987]

Anybody else noticed a memory leak on v5.8.0.5 ?

BI is currently using 22gb of RAM and counting. It crashed completely yesterday and i suspect the same will happen again

[compjunky]

^^^ All the yes.

Just upgraded last night and got an alert in the middle of the night that my Blue Iris machine was low on available RAM. When I logged into the machine and Blue Iris was using 30GB of RAM.

Restarted the service and it went back to using the normal amount of RAM it always has -- but it just keeps creeping back up. Looks like I'm going to need to roll back to the previous version until that is resolved.

Support data (sanatized):
Version: Release 5.8.0.5 x64 (10/12/2023)
Service: Yes [LocalSystem]
OS: Windows 10 Pro
CPU,GPU: Intel(R) Xeon(R) CPU X5680 @ 3.33GHz [49%,0%]

[awitty987]

Thanks, glad it's just not me. I think i'll also downgrade.

[compjunky]

Nope, not just you. I did send their support an email as well. So hopefully they'll catch it and fix it with the next release.

[louyo]

Interesting, I just updated this morning. I don't have the console open, only running as a service. It has been sitting on 180 meg for quite a while. I only have 3 cameras on that system and AI is on a different system.

[TimG]

Good catch. Mine usually uses around 1GB RAM, but was at 2GB and climbing slowly since I updated to it this morning. v5.8.0.4 it is then.

[Swanny_221]

I just started a thread on this but for me it has been an ongoing issue for some time, not related to a recent update. I'm currently on 5.8.04 but I've been seeing an intermittent problem for months. See my post about it being potentially storage related.

[infecticide]

I can confirm I have the same issue.

- Windows 11 VM on Proxmox
- 10 Cameras (mix of 3MP all the way to 4K)
- All recording direct to disc at full resolution
- Clips limited to 1Hr
- My BI instance averages 2.8GB - 3.5GB of RAM
- Deepstack is running elsewhere
- CPU usage is around 40% avg

BI 5.8.0.5 will run for anywhere from 4 to 12 hours before memory usage sky rockets suddenly (by 1 - 2GB / second) and takes the VM with it.

I have tried:
- Rolling back Windows Updates
- Deleting all BI storage and starting over fresh
- Rebooting all cameras

I have rolled back to 5.7.9.12 and its been running fine since, now going on 26 hours.

[Swanny_221]

I can confirm I have the same issue.

- Windows 11 VM on Proxmox
- 10 Cameras (mix of 3MP all the way to 4K)
- All recording direct to disc at full resolution
- Clips limited to 1Hr
- My BI instance averages 2.8GB - 3.5GB of RAM
- Deepstack is running elsewhere
- CPU usage is around 40% avg

BI 5.8.0.5 will run for anywhere from 4 to 12 hours before memory usage sky rockets suddenly (by 1 - 2GB / second) and takes the VM with it.

I have tried:
- Rolling back Windows Updates
- Deleting all BI storage and starting over fresh
- Rebooting all cameras

I have rolled back to 5.7.9.12 and its been running fine since, now going on 26 hours.

As a follow-up question, after deleting all BI storage, did you see any reprieve or did the memory usage immediately start increasing again? I ask because on my system (still 5.8.04) the memory leak stopped after the larger of my two storage drives got wiped due to a strange glitch (see my thread for details), but I'm expecting the problem to start up again once I reach my standard 14 day storage limit or the allocated space limit, at which point BI will have to actively prune things back. I believe in at least my own case there is a glitch relating to this process, though my smaller drive is still "at capacity" and I'm not seeing the issue.

[infecticide]

The issue continued, possibly with a bit more time between failures.

I was muddying the waters a bit by changing other things around that time.

As of now I've still not seen a failure after going back to 5.7.