Closed RobertHerter closed 5 years ago
@mhilbush Perhaps you can add a summary of your findings in #861. Perhaps another @openhab/core-maintainers likes it, too and merges the changes. I don't see any critical blocker, but that is normal for the one that submit a change request. So, I am interested in feedback (but perhaps #861 related topics should be added to that topic).
Perhaps you can add a summary of your findings in #861
Certainly! I'll do that now.
Maybe just a side issue, but with snapshot (1607) as well as my productive environment (2.4) ESH-usb-serial-discovery-linux-sysfs-1 takes a lot of load, though I do not use any binding using any usb or serial device. I see this service in most of your debugging taking much load either and being the second most consumer after #861 is solved.
@bjoernbrings By default, it polls every 5 seconds (which seems a bit aggressive to me). Fortunately, the thoughtful developer made it configurable. 😄 I'm going to try changing it in runtime.cfg
to every 5 minutes to see what happens.
discovery.usbserial.linuxsysfs.pollingscanner:pauseBetweenScansInSeconds = 600
Edit: Sure enough, if you want to tame ESH-usb-serial-discovery-linux-sysfs-1
, just put the above line in services/runtime.cfg
with whatever pause amount you want.
Just in case you need another benchmark, I'm running 2.5.0~S1601-1 (Build 1601) on a RPi 3B+ running OH for almost 15 days (1296000 seconds) now.
Here's an overview of the most resource-hungry processes (build 1601).
Id | Name | State | CPU time | Usr time |
---|---|---|---|---|
61 | ESH-OSGiEventManager | TIMED_WAITING | 90318011 | 89700470 |
227 | ESH-usb-serial-discovery-linux-sysfs-1 | TIMED_WAITING | 7184608 | 4904440 |
110 | pool-5-thread-1 | WAITING | 2681233 | 2419920 |
88854 | SocketListener(JmDNS-/192.168.0.109) | RUNNABLE | 2600052 | 2445420 |
179 | Thread-80 | TIMED_WAITING | 2138434 | 2101060 |
88848 | SocketListener(JmDNS-/192.168.0.10) | RUNNABLE | 1815205 | 1689120 |
88851 | SocketListener(JmDNS-/2a02:1810:481f:dc00:db4d:5:5c4d:941a%wlan0) | RUNNABLE | 1277256 | 1190630 |
88845 | SocketListener(JmDNS-/2a02:1810:481f:dc00:8e8c:3190:582a:64a7%eth0) | RUNNABLE | 1270393 | 1184900 |
203 | pool-26-thread-1 | TIMED_WAITING | 1083512 | 954290 |
240 | ZWaveReceiveInputThread | RUNNABLE | 1079327 | 810160 |
109 | pool-1-thread-1 | WAITING | 963790 | 718210 |
130 | openHAB-job-scheduler_Worker-2 | TIMED_WAITING | 759120 | 693390 |
88855 | JmDNS(JmDNS-/192.168.0.109).Timer | TIMED_WAITING | 736947 | 686500 |
129 | openHAB-job-scheduler_Worker-1 | TIMED_WAITING | 722831 | 659880 |
32 | EventAdminAsyncThread #11 | WAITING | 470794 | 419450 |
33 | EventAdminAsyncThread #12 | WAITING | 469847 | 418680 |
59 | EMF Reference Cleaner | WAITING | 461880 | 239760 |
28 | fileinstall-/var/lib/openhab2/etc | TIMED_WAITING | 457392 | 429390 |
88849 | JmDNS(JmDNS-/192.168.0.10).Timer | TIMED_WAITING | 418765 | 367320 |
183 | DTLS-Receiver-0.0.0.0/0.0.0.0:50954 | RUNNABLE | 390140 | 356660 |
13 | Karaf Lock Monitor Thread | TIMED_WAITING | 379288 | 282240 |
242 | ZWaveReceiveProcessorThread | WAITING | 354837 | 336360 |
88846 | JmDNS(JmDNS-/2a02:1810:481f:dc00:8e8c:3190:582a:64a7%eth0).Timer | TIMED_WAITING | 327078 | 285380 |
88852 | JmDNS(JmDNS-/2a02:1810:481f:dc00:db4d:5:5c4d:941a%wlan0).Timer | TIMED_WAITING | 322746 | 284050 |
123 | JmDNS(JmDNS-/192.168.0.109).Timer | TIMED_WAITING | 308996 | 287680 |
181 | DTLS-Sender-0.0.0.0/0.0.0.0:50954 | WAITING | 296296 | 231650 |
37 | fileinstall-/usr/share/openhab2/addons | TIMED_WAITING | 237867 | 208590 |
8 | Active Thread: Equinox Container: 5c8af011-8fc3-4b72-bca2-83f31477445e | TIMED_WAITING | 160843 | 137070 |
55 | sshd-SshServer[c1d5e6]-timer-thread-1 | TIMED_WAITING | 113117 | 80070 |
@maggu2810 , should this be closed now that #861 is merged, or are you waiting for some more feedback?
IMHO it can be closed as the issue should be resolved. @RobertHerter if you don't agree because the issue is still present in the next build, please comment.
FTR, I confirmed the performance of build 1618 matches my testing with #763 and #861.
For me issue is home with the new build. Thank you for the hard work. So at the Moment only Rest-Doc is not working in 2.5 everything working.
Since Build 1579 there is extreme high CPU-Load and Memory consumption.
OSGIEventManager uses nearly tue complete CPU. Reverting to 1578 instantly fixes this.
Even a Fresh install ends up in this, weit no Bundle found causing this.
Testet on an I5 and RPI3 both show same behaviour resulting in an nearly unusable system.