Probable memory leak

[trevor-austin]

Ever since I installed Jowihue, currently running version 1.1.2.6 I have noticed that I need to reboot my HS3 server every few days to keep it running reasonably.

I have monitored it for the last few days and the jowihue process gradually consumes more and more memory and CPU cycles until HS3 stops responding. I am running on 2012 VM with a lot of memory and processing power.

I just disabled and renabled the add in through HS3 and it went from 25% CPU and 741,02KkB of ram in use to 0-2% CPU and 41KB of ram ( but growing at 20-60K every few seconds)

[w.vuyk]

Trevor,

I have checked the HSPI_JowiHue proces on my system just now (Windows, low capacity CPU en 2Gb memory) and it is using 16Mb memory/0% CPU. This is for a dual bridge/24 lights/8 sensors configuration
HS3 is running now for 11 days. HS3 itself is running at 300 Mb/5% CPU.

So I cannot see the issue from here. Do you have multiple bridges/sensors or a lot of lights maybe? Or anything other special?

Wim

[trevor-austin]

Trevor,

I have checked the HSPI_JowiHue proces on my system just now (Windows, low capacity CPU en 2Gb memory) and it is using 16Mb memory/0% CPU. This is for a dual bridge/24 lights/8 sensors configuration
HS3 is running now for 11 days. HS3 itself is running at 300 Mb/5% CPU.

So I cannot see the issue from here. Do you have multiple bridges/sensors or a lot of lights maybe? Or anything other special?

Wim

Hi Wim,

I have one live bridge 2, and an old bridge 1, but that is turned off. I do also use the HA-Bridge tool, and a number of devices controlled by it and yes quite a few lights, 21 in total.

I just checked and it's currently stuck at 25% CPU, sometimes dropping to 10-13, and 426,158kb and growing memory use.

I killed off the HA-Bridge service but it made no difference.

[w.vuyk]

Hi Wim,

I have one live bridge 2, and an old bridge 1, but that is turned off. I do also use the HA-Bridge tool, and a number of devices controlled by it and yes quite a few lights, 21 in total.

I just checked and it's currently stuck at 25% CPU, sometimes dropping to 10-13, and 426,158kb and growing memory use.

I killed off the HA-Bridge service but it made no difference.

So the number of lights is similar to my situation. As for the HA bridge, this is not in my configuration. In the new version that should be released by HST any day now, the plugin will recognize the fake bridge and ignore it. Maybe this could help your issue? Let me know once the version is available?

Just to be sure, you mention 25% CPU, is this the HS3 process or the HSPI_JowiHue process? Same question for the memory?

Thanks,

Wim

[trevor-austin]

So the number of lights is similar to my situation. As for the HA bridge, this is not in my configuration. In the new version that should be released by HST any day now, the plugin will recognize the fake bridge and ignore it. Maybe this could help your issue? Let me know once the version is available?

Just to be sure, you mention 25% CPU, is this the HS3 process or the HSPI_JowiHue process? Same question for the memory?

Thanks,

Wim

Hi,

Those figures are just for the HSPI_Jowihue, the HS3 process is 92mb and 0 CPU while doing nothing as expected, i rebootted at 8:30 am this morning and the memory/cpu use is gradually climbing back to the 25% CPU (seems max, as in maybe 1 full core, is it multi-threaded?) and memory just seems to grow constantly until it falls over and I get an error saying Jowihue has stopped working. I have had to schedule a VM reboot every 3 days to keep things running.

This is current CPU use for all HSPI devices after 90 minutes uptime.

[w.vuyk]

Trevor,

Sorry for the late response here. In the past period I have tried testing on a larger computer, to see if this would show anything different from running the plugin on a small system, but I am not able to replicate your issue here. That makes it quite hard to analyze what is happening on your system.
I don't know what really is happening for you that is causing this issue.

Could you do a trace maybe, hoping that this would give some insight on what is happening? Make sure you are on the last version of the plugin and follow the next steps for the trace:

• On the configuration page of the plugin enable deeptrace en trace to file options
• Stop the plugin and restart it, so I will hav a trace from the start
• let the plugin run for a while until you see the CPU usage growing for some time
• then disable both deep logging and trace to file.
• No need to restart the plugin, get the JowiHue.log from the logs subdirectory under the HS3 installation folder
• Please zip this log and send it to me (w dot vuyk at upcmail dot nl)

I hope I will then be able to see why the plugin is getting nervous on your system?

Thanks,

Wim