CPU usage increasing after a few hours #158
Comments
|
During a restart, there are a couple of maintainance things to do and can take some minutes. After that there is just one thread and asyncron calls, so basically if you had 20% CPU before you restart the logger, it would help me to see in what state of the statemachine it was, so I can search for potentially infinite loop. could you please send me the whole logfile? It is located in /etc/teslalogger/nohup.out |
|
Hi, |
|
I don't know what happen to your Docker, but the time seems to jump back in past. I really cant say, what teslalogger will do, if that happen. 27.03.2020 18:24:17 : GeocodeCache:Insert 27.03.2020 18:18:46 : Docker detected |
|
That's strange indeed. I will check this also. |
|
I monitored the CPU usage in continuous and CPU usage increase is... continuous: see attached graph |
|
|
|
Did you make a reboot on 30 March. I'm asking because of the drop. I'll ask the other Docker users if they have the same problem. |
|
Yes I did a restart of the teslalogger container |
|
I tried to run the docker on my synology NAS Docker, but after a couple of days the cpu was at 5%. I try to figure out, what the difference is between regular install and docker. |
|
same here :/ |
|
|
|
I can look into my docker installation, what software do you use to get the graphs and info so that I can present same info as you do |
|
I use monitorix, really easy to use as a docker container (erikmroll/monitorix) |
|
I have installed it now, just waiting to collect some data now. Will get back when I have some more to show |
|
after 25h 1129MB RAM?! |
|
Docker on Synology running 3 Days. Just 69MB Ram and 3% CPU.
|
|
been running Monitorix since 09:00 this morning and can't see anything that's unusual.
|
|
@DariBer Your CPU usage has increased, going from almost nothing to 5%... Let's see if it increases more. The absolute value of the CPU usage should not be compared between users. My docker containers are running in a VM on an ESXI server and this VM is assigned 1 core of an Intel(R) Core(TM) i3-3250 CPU @ 3.50GHz. If you run your docker with 4 more powerful cores, the CPU usage will maybe reach 5% when mine may be at 20%... But you you see the trend anyway: CPU usage will increase.
|
|
It has soon been up for 24h and I had a peak at 8% the average is 6,8% and current is 3,6% |
|
@DariBer What hardware does it run on? |
|
I turned on long term cpu statistics on my Synology and no increasing of cpu is visible.
|
|
I did a lot of optimisations in "sleep" mode, so I drop the CPU on my Synology from 5% to 0,5%.
Changeset: |
|
Please update to 1.40.3.0 |
|
Nice optimisation indeed! |
|
That's better, but still not ok: usage goes from 0.5% to 10% in one day. |
|
|
|
RasPi, 1.40.3.0: |
|
|
bassmaster187
changed the title
|
I could rule out, there is a bug in the code with profiling the code over 8 hours with all states except "unknown" and "offline".
Some users reported also high CPU usage in Raspberry, so i changed the topic of this issue. So I started to compare my Raspberry Image with my Synology Docker. Same config, same car, same credentials. Raspberry has 0% avg, Docker 5% after one day.
So the only difference between my enviroment is Mono Version. I'll try to use a newer mono version in docker. |
|
Tried some settings that might be relevant for FileSystemwatcher ala both both had no effect after ~12h mono is up to approx 30%CPU again on a RPI3b with latest mono installed. |
there is no screenshot in your reply |
|
2nd try with an attachement. 1st try was an inplace picture in an html
email.
Am 29/04/2020 um 09:09 schrieb Christian P.:
…
Here you go: Am 28/04/2020 um 16:31 schrieb Christian P.:
… <#>
I have an updated "Raspberry Pi 3B" (with latest Mono), so could
be interessting to cross check with a docker image. If you like me
to test something, I happy to do so. Just tell a windows guy what
todo - I am used to managed my Windows infrastructure using
command line, so any SSH commands and/or sudo stuff is no issue
;-) just need the details what to do. please make a screenshot of
"top" command in ssh. I need the uptime (top left corner) and
"time+" of the mono. — You are receiving this because you
commented. Reply to this email directly, view it on GitHub <#158
(comment)
<#158 (comment)>>,
or unsubscribe
https://github.com/notifications/unsubscribe-auth/AL4L3BBWOUS6AD72WHUPSDDRO3SCXANCNFSM4LVS7IXA.
there is no screenshot in your reply
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#158 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AL4L3BDEJSOH6JWZEXN4GQDRO7HD7ANCNFSM4LVS7IXA>.
|
|
I also noticed the cpu load increase using docker. Started monitoring the mono process now. One sample every 60 sec with average over the last 10 sec. |
|
Please update to the latest version. I made a lot of performance impovements and caching. CPU is below 1% and I hope it's not raising.
|
|
I upgraded to the latest version and will continue to monitor. |
|
Not OK yet :-(
|
|
Try changing mono version for teslalogger container to 5.20.1.34. It worked for me. |
|
On Raspberry 4 with Mono 5.8.0.240 everything is ok:
So I think it is caused by Mono V6 |
|
Compatibility Issue with Mono 6, was confirmed now by minimum 3 users. |
|
It really seems like it is caused by Mono 6. After 14 hours with Mono 5.4.1.34 with Docker on my Synology I'm way below 1%
Is anybody out there with Mono 5.4.1.34 with increasing CPU usage? I think I'm gonna open an issue in Mono Repository. |
|
Opening an issue is certainly a good idea. However in the meantime you could maybe use a previous version of mono in the container? |
bassmaster187
changed the title
|
Now running downgraded mono version... |
|
|
|
Same here, so I think we could close this issue and Mono V6 is causing this Problem.
|
|
Has this issue ever been solved? I`m running teslalogger with Docker on a Lenovo ThinkCentre M900 an Teslalogger/Mono eats up 1 CPU core permanently. Mono Version: 6.8.0.105
|
|
No Problems with Mono 6.12.0.182
As far as I can remember, the problem was creating a new WebClient caused this problem. Now we are reusing the WebClient Class. |
|
Updated the version of Mono to the latest 6.12.0.200, sadly the problem still persists. A few seconds after the start of Teslalogger it eats up 100% of one core. |
|
After the start of Teslalogger it is common to use a lot of CPU to update the database and do some housekeeping. But if after lets say 1 hour the problem still persists, we have a problem. Please run Teslalogger for at least one day. Get the PID of teslalogger: Use the PID in following command: Post a screenshot like this: |
|
I let it run for about one day, but it stays at 100%. Here´s the output you requested: |
|
Looks like it's generating maps for the timeline. |
|
Could you please post a logfile, so we can see if it's a endless loop or really generating maps for the timeline |
|
Logfile Teil 1 |
|
Logfile Teil 2 |
Did you see my post? |
I run the lastest version of Tesla Logger in docker (updated from the admin page) and every time, after a few hours, the teslalogger container starts using cpu constantly (20% of my cpu).
If I restart it, it drops to less than 5% until some unknown problem that causes it to use more CPU. The log doesn't give any information when this triggers.
Other containers don't use more than a few % CPU.
Car Version: | 2020.8.2
Last Update: | 27/03/2020 à 19:18:56
Teslalogger: | 1.39.0.0
The text was updated successfully, but these errors were encountered: