Memory Vera plus

As far as I remember, if Vera is unplugged from internet it may affect its performance (I think I’ve read some threads about that here), so unplugging may be not a good idea.
Some routers have ability to create separate guest network, you may try that way.

Yes, so probably removing all and installing them back one after another is better solution than the other way around. Still, lot of work and time. Especially for PLEG actions where I wouldn’t re-create them all at once, but also one by another.
From the other hand, leaving your system as it is now is in my opinion unknown: it can either work for years like that, or you may get major breakdown soon.

And yes, you have to remove plugins, my mental shorcut about “disabling”

yes… unplugging the vera will make it unhappy. If your router does not support separate networks/vlan/subnets, then one option is to put the vera behind a different router where its WAN interface is plugged in to a LAN interface on your main router.

Another thing to consider that just uninstalling the plugin does not remove the files, and I suspect the vera still loads them in to memory in manner (or at least in the past). For some suspect plugins, I would delete the files manually in the /etc/cmh-ludl directory.

One more thing, I had to recently remove DataMine from my Vera as tech support said it was causing problems with space exhaustion before an upgrade. This was a storage issue and not a memory problem that I know of, however I just decided to remove DM to avoid the issue in the future.

@ Niharmehta: Vera was acting the same, only when the vera reboots it needs the internet for the current time.

Was your datamine data on an USB stick? could that have been the issue, mine all data goes onto an USB stick.

Cor

Originally it was on a USB stick. It has been a while, but I think I had combination of issues with a bad USB stick, and DM taking too much space on the /www directory … I forget the details now of which combination of the two were my main issues.

Over the years, I have gone through similar issues as you have. Fully deleting plugins via clearing the /etc/cmh-ludl directory and isolating the vera on its own broadcast domain seemed to have stabilized it. I have not had an unplanned reboot in many months and have a relatively loaded system. Hopefully you are close to isolating your cause.

I can’t find it now, but there was a topic where someone had issues with Vera due to lack of internet connection, and not necessarily it was about reboot and setting time.
Remember that part of your plugins need internet connection as well and lack of it may affect your system stability even if it don’t affect the core system directly.

Datamine is tricky in setting and sometimes it doesn’t “catch” proper settings immediately. I had an issue where I set it correctly to store data on USB stick (or so I thought), but in fact it was stuffing data into local memory. I don’t remember how exactly I’ve solved this issue, I think I formatted the stick and made whole configuration again.
If you set Vera to save its log into internal memory it may also cause a problem.

After I made sure to disable vera local logs and that DM is saving data on the stic, I don’t have issues with memory in general.
Nevertheless I monitor both, internal and USB stick memory.

2 weeks ago I asked help from the vera techs, they came with this answer:

I have checked the logs and indeed the free memory of your controller goes up and down. Most likely, this issue is created by one or more of your plugins. However, it is hard to identify the exact plugin(s) as all the running processes are stored under the same name ? LuaUPnP. You can use the following workaround to force the unit to clear up the cache memory:
  • create a scene that will run every 30 minutes and add the following luup code :

os.execute(“echo 3 > /proc/sys/vm/drop_caches”)
Let me know how it goes.

Today I had finally some time to look at my system again and I inserted this code in a scene, running every 30 mins:

os.execute("echo 3 > /proc/sys/vm/drop_caches")

It looks like this does the trick (see attached screenshot, as from ca 10:00 ) , the line a quite flat now , compared with the previous days.
Although I do understand , this is not a solution to the “problem” it looks like it is a good work around.

To remove all the plugins , that would be not so good , and would take a huge amount of time , also when I find a plugin which is the problem , it would mean , I am unable to use it , meaning; my vera would be quite useless.

Cor

Thanks so much for reporting this.

[quote=“Cor, post:27, topic:198791”]You can use the following workaround to force the unit to clear up the cache memory:

  • create a scene that will run every 30 minutes and add the following luup code :

os.execute(“echo 3 > /proc/sys/vm/drop_caches”)[/quote]

Definitely worth a try.

It’s also something to wonder at when they say:

all the running processes are stored under the same name ? LuaUPnP.

or, in other words:

,,,we know this is bad design and makes it hard to trouble-shoot, but we've never got around to doing anything about it.

…useless!

Yep , indeed useless.
If I would have known before when I started this Home automation journey , I should have started with Homeseeer. Anyway I am into vera now and everything is working …most of the times.

Anyway , I think that code:

os.execute("echo 3 > /proc/sys/vm/drop_caches")

is doing a great job on my system. Attached another screenshot. The system is now running around 30 hours without issues. And the memory lines are more or less flat.

Cor

Read the comment here about this script:

http://forum.micasaverde.com/index.php/topic,60670.msg352341.html#msg352341

I’m not competent enough to comment on this. I use this script to flush memory in case of significant drop of memory, but in my case it is once a month or even less

…and here’s one of mine. Can you see where it started?!

The system is now running around 30 hours without issues. And the memory lines are more or less flat.

…still doing well?