So it’s been about a week since I have posted.
I still get the crashes… anyone else?
Has this topic died?
Any new news from anyone? This doesn’t seem to spark much interest (or support), so I’m guessing it isn’t all that important?
Kind of frustrating to continue to have my display crash on multiple OS platforms possibly due to memory leaks.
When I run OpenEVSE, I don’t have this problem.
My weather station doesn’t have this problem.
My security system (BlueIris) doesn’t have this problem.
Openenergymonitor gauges and/or graphs DOES have this problem.
Kinda points to the webpage and how things are being done, and not my system(s)!
Rather than not being important, it seems to have been difficult to replicate the issue. @TrystanLea is the best person to help but he has other things on the boil and until he can replicate the issue, it is difficult to debug.
I suspect you are right. You say you have a dashboard of just dials/guages - are you able to pin down a bit more what might be causing the issue? Does just one of the widgets cause it? Does it matter what the feed engine is? You have eliminated different platforms and OS so it could be the widget in some way.
I have no idea how to determine which widget might be/is causing the crash.
I have created a dashboard with only one widget (a dial) that is displaying the voltage.
I will let that run and see if it crashes, and if it does I will create another dashboard with only one gauge and let that run.
This should determine if the dial or gauge is causing a crash.
If neither dashboard crashes, I will add a couple more dials to the dial only dashboard and let that run.
I might even create a totally blank dashboard just as a test. I would think that none of the widgets / graphs are causing the crash. But as far as I could determine, that is above my experience.
Sorry for the late reply… family “interference” hehe.
Yes, I setup all my feeds to PHPFINA engine at 10 second update rate. That what was recommended.
Seems strange because even though they are set at 10 seconds, they seem to update at about 3-5 seconds.
I have run my single dial dashboard, single gauge dashboard for 24 hours and no crash.
I also have been running a 5 dial dashboard and so far it hasn’t crashed either.
If it survives more than 24 hours, I will create a 5 gauge dashboard and run that one.
Noting your post @pomonabill220, I’m still unable to replicate this here, but due to other work items going on in parallel I’ve only been able to try long term tests on a screen connected to a raspberrypi running chromium. Is it perhaps worth trying chromium?
I wasn’t monitoring memory usage during the run, darn it!
This run, with 5 gauges, I am, and ~12 hours later it is around 398Mb max.
This has happened before, and before the other dashboard(s) crashed, the max memory could climb to almost 2Gb.
I will continue to run this latest dashboard with the memory monitoring and IF it crashes I hope to capture the max memory usage.
I guess I could try another browser. Any tricks installing and using chromium to display the dashboards? It doesn’t look like anything “special” for configuration from what I can see.
My 5 gauge dashboard has been running non stop for 48 hours… NO crash.
I have also been watching memory usage and it varied between about 128Meg and 357Meg with a peak of 389Meg.
Can’t explain why the memory usage varies so much though. Nothing else is running on the tablet either.
I am going to try the 5 dial dashboard again, and watch memory usage this time.
Thanks @pomonabill220, I’ve renamed the topic just in case it catches the attention of anyone else experiencing the same. Also interested to hear from those running dashboards for extended periods of time without an issue?
I tried a test on an old android tablet using chrome a couple of days ago which appeared to continue to work ok. Im now away from the office until the end of the month so wont get a chance to continue testing until then.
My gauges dashboard finally crashed and the memory reported peaked at 0.92Gb!
WOW!
I started my 5 dial dashboard with an app that graphs memory usage (although not specific app usage) and it started at 52% at about 128Meg, and after ~12 hours it is up to 62% peaking at 357Meg.
I will continue to let it run and watch it.
IF no one else has this crashing problem on Android, running continuously, I would like to know how their system is setup, what dials/gauges/graphs they are using and how many, and how long they have been running them. How their feeds and inputs are configured? Any memory “cleaners” running? What OS are they running their dashboards on?
i was reading that Android can have a memory leak problem even though it does do memory cache maintenance, but cannot perform the maintenance if the memory is “locked” by an app
I am no programmer, but I do have some knowledge of the concepts used.
There may have been a few clues as to what is going on as the memory usage keeps bouncing around, but finally just climbs out of sight before a crash.
I tried closing and restarting Chrome/incognito mode, with the dashboard and the memory footprint started at the low value, so the memory footprint was cleaned up.
I have been running my 5 dial dashboard with a memory monitor running in split screen mode.
It started at about 1 gb usage (I believe that is total memory usage with this monitoring app)., then during the day it crept up to 1.3Gb. No crashing yet though.
Then, this morning the usage had dropped to about 900Meg!
There doesn’t seem to be any drastic peaks going on, just a very slow increase in usage.
I am going to let it run some more and see if the usage creeps up again.
I wish this app would allow individual process monitoring and logging, but it doesn’t seem to.
After I let this dashboard run for another, say, 12 hours (this evening), I am going to start my “default” dashboard (with all the graphs/gauges/dials… the one that reliably crashed), and keep an eye on the memory usage and how it acts.
Still stange that the memory footprint would creep up, then BAM drop off… hhmmmm.
I ended the 5 dial dashboard and the memory used topped out at 1 gig.
I started my default dashboard id=1 with gauges/dials/multigraphs and the memory started at 1.04 gigs.
As soon as I started the id=1 dashboard, the memory jumped to 1.04 Gigs also.
I will let it run and hopefully catch it before it crashes, or at least be able to see the memory used over time.
Let you all know what I find.
Here is something interesting that I caught today. The id=1 dashboard had been running for about 24 hours along with the memory monitor app, and used memory started at 940Meg, and throughout the day had crept up to a little over 1.4 Gig. THEN something “cleaned up” memory and the useage dropped to about 940Meg.
I did not do anything to cause this, but I took a screen capture of my graph and attached it here.
Each time division (vertical) is 20 seconds. (the max I could set)
Well after running the memory monitor and id=1 dashboard for 20 hours, it finally crashed.
The memory usage was only about 1 Gig with no creeping, so it looks like something else is causing the crash.
All I had to do to 'restart" the dashboard was to reload the page. and it is continuing with about 1 Gig of memory usage.
Any other ideas? I am fresh out myself… becoming frustrating!
Does anyone have a dashboard that has run on android in chrome or firefox, and has run for more than 24 hours? What are your settings and what are you displaying?
I would like to try your settings on my tablet just to see if it will run for me… kind of cross checking if you will.
