I tried that limiting, but on my system the patch had some negative effect. I guess I am already past the 650mb limit. 
I currently moved openHAB to my file server Core2Duo/4gb ram. openHAB uses 1.5gb and about 300 threads. So this explains somehow my sporadic lags in rule execution. Looks like it is time to move to another hardware platformâŠ
Huh?
Note the patch above only increases the static memory footprint (process size). Itâs up to the OS to select what of that memory to page out.
On startup that can result in some additional paging, but once you access a (compiled) rule that was paged out, it just needs to be paged back in, and that should be a lot faster than to recompile at that stage.
Note I am not using 650MB but just 400M on my 1GB Pi, that resulted in an increase to the static process size of ~130MB. Also, Iâm swapping (paging) to my NAS as described here.
Thereâs a thread somewhere people telling how large their Java process is, itâs usually around 650 MB as is mine.
If youâre at 1.5G it sounds like you need to investigate your system. It shouldnât use 1.5 G. Do you use any statically defined large objects in your rules ? Iâm referring some MP3 files, and Java sort of seems to pre-load them. At least I had choppy audio on those and the patch made that go away.
Or any binding to do some strange caching or to keep large objects such as database indices ? If so, try to reduce those settings.
Is it really 1.5G of resident memory use or just the process size?
And youâre still having issues with 4G ?? That should be more than enough.
I understood the patch sets the min and max memory of the java process to 650mb, effectively making it fixed size. What i discovered when using the patch is a slow and unresponsive BasicUI (no icons etc.).
This is my output from htop, with the patch removed.
According to my interpretation, my installation of openHAB uses 1.3 MB of RAM. (And about 300 threads, because the number of threads goes down by about 300, if i stop openHAB)
The audio files in the /etc/sounds folder are about 7 MB, icons are about 2.5 MB.
Looking at my memory, i would guess i should use larger numbers, 1500 instead of 650, to make this patch work for me currently.
1.3 GB resident indeed.
Are you running on ARM ? Which Java ?
And how much RAM does it use when started on your RPi ?
Try 3000m or even 4000m for the max value as it affects static process size and you already seem to use that much.
No, Amd64, Ubuntu 18.04.1 LTS, Core2Duo with 4 GB of Ram (Board Limit).
Itâs Oracle
java version "1.8.0_181"
Java(TM) SE Runtime Environment (build 1.8.0_181-b13)
Java HotSpot(TM) 64-Bit Server VM (build 25.181-b13, mixed mode)
I have to check that, i moved openHAB to the file server a few weeks ago, because of some extreme lagging of my motion sensors rule execution. Not nice if the light turns on 3 minutes after the motion is detected. 
I think i should check, how to reduce the memory footprint. I have about 150 things and about 2000 items. Is it possible to check the used memory per binding somehow?
Now sorry to tell but you sort of made a mess of this thread.
It was about running on RPis, and the memsize recommendations were specific to that (i.e. a box to have an ARM processor architecture and 1GB of RAM, with Zulu Java, usually resulting in ~650 MB process size).
But now you state you run on some 64bit x86 and Oracle Java which is totally different as itâll use 64bit and thus much more memory for the same Java code and openHAB config.
No wonder it broke your system when you applied that RPi specific memory optimization setting.
Do you mean to say that you still have problems even with that 4GB Intel machine ?
(btw, amd64 on Intel CPU does not sound right either).
Given your htop output of 3,8GB(!) Java uses, that HW/OS/Java combo possibly was a worse choice than a RP is as 3,8 out of 4 GB is a higher fraction than 650MB is of 1GB.
Either way, if you want to compare your figures to ours you should do this on the same processor+Java basis.
So start your config on your Pi again to see how much RAM it takes there.
Sorry, i will delete my posts if you prefer.
No worries. Iâm not sure whatâs up with his attitude, might be the language barrier. âGerman Englishâ can sound kinda rude 
Agreed with illnesse. Please leave your posts. But in the future when being helped to debug a problem, particularly one of this nature, you must tell those who are helping you EVERY modification you make, especially if you reinstall. Otherwise we are operating with incorrect information and assumptions.
Out of curiosity, you are not running in Docker, right? Because when running in Docker the amount of RAM returned is the total for the container, not just the OH process, which can be significantly higher.
Minimal OH installed on an RPi
Modest OH configuration running in Docker
A short summarize of my origin problem: I only make modifications âofflineâ and only load up file by file to the rpi, so I can check after every upload if the changes work. The system works better as before with OH2.2, but itâs still not as fast as OH 2.1. I think, this is what we have to deal with.
The contra of this technique is: while programming offline, thereâs no error-highlighting in VS-Code, I think. So, I really have to be careful while writing. Eclipse Smart Home Designer does not work on my Mac. Is there an other possibility to find out errors in syntax ?
The only syntax checking available is VSCode which requires a connection to a running OH or saving the file in the appropriate folder and watching openhab.log for errors.
However, you do not need to be editing the files on the RPi where OH is running to get syntax checking. You just need to have a network connection to a running OH server. So you can edit the files on another machine and move them over as desired.
Hardware: RPi 3B+
OS: Openhabian
Java Runtime Environment:
openjdk version â1.8.0_202â
OpenJDK Runtime Environment (Zulu8.36.0.152-CA-linux_aarch32hf) (build 1.8.0_202-b152)
OpenJDK Client VM (Zulu8.36.0.152-CA-linux_aarch32hf) (build 25.202-b152, mixed mode, Evaluation)
openHAB version: Was 2.4 Stable and upgraded to 2.5 M1 (read inside)
My system includes Z-Wave modules
All of my Things were created using Paper UI
All of my Items are defined in files
All of my rules are defined in file
Using SSD drive (no SD card at all)
My experience with OpenHAB started with version 2.4 Stable on a RPi 3B+ with Openhabian.
I have always been using VS on a Windows machine to directly edit the rule files, which resides on the RPi.
I was using DIRECT saves and SAMBA, both of which (I found out 2 days ago) are not recommended.
In any case, even though I was working this way, whenever I saved the changes I made to a specific rules file, it was immediately ârefreshedâ, and after a few seconds I could already see the new logic in action.
About 2 days ago, it was no longer the case.
After I saved some changes, the system stopped running all my rules (later on I understood it probably just became very slow).
Stopping and starting the openhab2 service (including clearing the cache) did not solve the issue (the system was slow even after restart), and I also came across the following error:
Dispatching event to subscriber 'org.eclipse.smarthome.io.monitor.internal.EventLogger@138f5b7' takes more than 5000ms.
After reading this, I upgraded to OH 2.5 M1, using the instructions here.
After that, when starting the openhab2 service, it works well (or at least it seems to).
I did come across these while writing this message:
2019-06-10 21:06:28.906 [ERROR] [org.eclipse.smarthome.core ] - bundle org.eclipse.smarthome.core:0.11.0.oh250M1 (102)[org.eclipse.smarthome.core.internal.scheduler.PeriodicSchedulerImpl(106)] : doFindMethod: Suitable but non-accessible method found in class org.eclipse.smarthome.core.internal.scheduler.PeriodicSchedulerImpl
2019-06-10 21:06:28.910 [ERROR] [org.eclipse.smarthome.core ] - bundle org.eclipse.smarthome.core:0.11.0.oh250M1 (102)[org.eclipse.smarthome.core.internal.scheduler.PeriodicSchedulerImpl(106)] : findMethod: Suitable but non-accessible method setScheduler found in class org.eclipse.smarthome.core.internal.scheduler.PeriodicSchedulerImpl, subclass of org.eclipse.smarthome.core.internal.scheduler.PeriodicSchedulerImpl
2019-06-10 21:06:28.914 [ERROR] [org.eclipse.smarthome.core ] - bundle org.eclipse.smarthome.core:0.11.0.oh250M1 (102)[org.eclipse.smarthome.core.internal.scheduler.PeriodicSchedulerImpl(106)] : bind method [setScheduler] not found; Component will fail
2019-06-10 21:06:28.917 [ERROR] [org.eclipse.smarthome.core ] - bundle org.eclipse.smarthome.core:0.11.0.oh250M1 (102)[org.eclipse.smarthome.core.internal.scheduler.PeriodicSchedulerImpl(106)] : bind method [setScheduler] not found
2019-06-10 21:06:28.921 [ERROR] [org.eclipse.smarthome.core ] - bundle org.eclipse.smarthome.core:0.11.0.oh250M1 (102)[org.eclipse.smarthome.core.internal.scheduler.PeriodicSchedulerImpl(106)] : doFindMethod: Suitable but non-accessible method found in class org.eclipse.smarthome.core.internal.scheduler.PeriodicSchedulerImpl
2019-06-10 21:06:28.923 [ERROR] [org.eclipse.smarthome.core ] - bundle org.eclipse.smarthome.core:0.11.0.oh250M1 (102)[org.eclipse.smarthome.core.internal.scheduler.PeriodicSchedulerImpl(106)] : findMethod: Suitable but non-accessible method unsetScheduler found in class org.eclipse.smarthome.core.internal.scheduler.PeriodicSchedulerImpl, subclass of org.eclipse.smarthome.core.internal.scheduler.PeriodicSchedulerImpl
2019-06-10 21:06:28.926 [ERROR] [org.eclipse.smarthome.core ] - bundle org.eclipse.smarthome.core:0.11.0.oh250M1 (102)[org.eclipse.smarthome.core.internal.scheduler.PeriodicSchedulerImpl(106)] : unbind method [unsetScheduler] not found; Component will fail
2019-06-10 21:06:28.929 [ERROR] [org.eclipse.smarthome.core ] - bundle org.eclipse.smarthome.core:0.11.0.oh250M1 (102)[org.eclipse.smarthome.core.internal.scheduler.PeriodicSchedulerImpl(106)] : unbind method [unsetScheduler] not found
But I believe these are a known issue, and are not related to my issue.
Still, whenever I change one of the rules files, it becomes slow.
I stopped editing the files directly on OH, and instead I upload them using MobaXTerm.
Sometimes when uploading the files, I receive the following warning:
[WARN ] [el.core.internal.ModelRepositoryImpl] - Configuration model 'filename.rules' is either empty or cannot be parsed correctly!
Sometimes, after this warning, the file is being re-loaded, and sometimes not.
This made me also try to upload the new rules files as
filename.rules.new
and then use the following command to rename it:
mv filename.rules.new filename.rules
I am not sure if it important or not, but the OWNER of my rules files is openhabian and the GROUP is openhab.
I modified it using:
sudo chown -R openhabian:openhab /etc/openhab2/rules
This is because when the owner was openhab I could not upload the files using MobaXTerm.
Still after working this way, sometimes the new rules file is not refreshed again, after the âis either empty or cannot be parsed correctly!â error.
Just to demonstrate how the system becomes slow, I have a Test switch item:
Switch ivTest_Rule_Trigger "Test Rule Trigger" {expire="2s,command=OFF"}
I created a test.rules file with the following content:
rule TestRule
when
Item ivTest_Rule_Trigger changed to ON
then
logInfo("Test Rule", "New Test 1")
end
Whenever I click the Test switch on Basic UI, all rule steps are executed very quickly:
2019-06-10 20:22:46.694 [ome.event.ItemCommandEvent] - Item 'ivTest_Rule_Trigger' received command ON
2019-06-10 20:22:46.717 [vent.ItemStateChangedEvent] - ivTest_Rule_Trigger changed from OFF to ON
2019-06-10 20:22:46.770 [INFO ] [pse.smarthome.model.script.Test Rule] - New Test 1
2019-06-10 20:22:49.663 [ome.event.ItemCommandEvent] - Item 'ivTest_Rule_Trigger' received command OFF
2019-06-10 20:22:49.681 [vent.ItemStateChangedEvent] - ivTest_Rule_Trigger changed from ON to OFF
However, once I update one of the other rules files (uploading a new version of it), it takes much longer for the logInfo comment to be executed:
2019-06-10 20:25:43.261 [ome.event.ItemCommandEvent] - Item 'ivTest_Rule_Trigger' received command ON
2019-06-10 20:25:43.270 [vent.ItemStateChangedEvent] - ivTest_Rule_Trigger changed from OFF to ON
2019-06-10 20:25:48.969 [ome.event.ItemCommandEvent] - Item 'ivTest_Rule_Trigger' received command OFF
2019-06-10 20:25:48.998 [vent.ItemStateChangedEvent] - ivTest_Rule_Trigger changed from ON to OFF
2019-06-10 20:27:47.160 [INFO ] [pse.smarthome.model.script.Test Rule] - New Test 1
(this is a 2-minute delay example, but I also had an almost-5-minute delay example)
Looking at the output of the TOP command, it seems like it is using high CPU, after uploading an updated rules filed (even a few minutes later):
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
32231 openhab 20 0 657600 513284 13932 S 139.1 51.4 135:19.98 java
Executing the commands listed here, it seems like I have:
41 Things
151 Items
139 Rules
2503 lines of rules
(note that since Iâm learning OpenHAB, I have MANY commented-out lines, that I removed after making all kinds of tests during my learning)
Iâd be happy to share any additional data that may be required to analyze the root cause of this issue, and see if it is a problem on my side, or a bug that has to be resolved.
This can be safely ignored. When saving the file what happens is there are two file events, one when it is is opened for writing and one when it is done being written. The warning you see for this is happening on that first event as OH tries to read the file before itâs been written to.
As long as the openhab group has read write permissions on the files it should be OK.
This is not an unexpected amount of a delay for an RPi. Iâve seen delays reported of over half an hour. That is why Markus in the post you linked to recommends against editing files using direct saves. Every file change results in a new parsing activity and if you donât wait long enough between file saves they pile up.
Thanks for your response.
But why sometimes there is no 2nd ârefreshingâ message?
I also just noticed that I continue receiving the error:
Dispatching event to subscriber ....... takes more than 5000ms.
even with OH2.5M1.
Also, my main question is:
How come I didnât come across any of these issues for several weeks, and only when editing a single rules file (with minor modifications) this whole slow-down issue comes up, and remains even after upgrading and after switching to uploading files instead of saving files?
By the way, when I stop and start the openhab2 service, it takes about 90 seconds for everything to start, load and work flawlessly (until I upload a new rules file).
(not taking into account the time it takes to stop the service)
Because there is something else going wrong. Either the file is in fact empty or unparsable or everything has slowed down to the point where it can be some time before the âloadedâ message appears.
Unless and until someone else experiences the same behavior and has the ability to investigate and diagnose the problem, no one can answer this question.
Will anything I do upload can assist you / the experts to investigate and diagnose the problem?
There probably isnât anything because someone will need to hook up a profiler to a running instance of OH exhibiting this behavior to see what parts are consuming all the resources.
And to be honest, you really havenât convinced me that the problem (and therefore the work around) is not as Markus describes.
I am not sure I understand this comment.
Do you refer to Markus mentioning NOT to work with direct editing and SAMBA?
If so - then since 2 days ago I STOPPED working this way.
I only edit the files locally on a different machine, and then UPLOAD it to the RPi.
Isnât this what I should do?
Or is there something else I should try?
OK, that wasnât clear. But if you are copying all the edits over as a batch and not one at a time, and giving plenty of time for the Rules to be parsed between changes then yes, it probably is a different issue.
But without being able to reproduce the problem I donât think it will be possible to provide much in terms of additional support.
All I can recommend is to make sure you are not using primitives in your rules which are known to slow down the parsing on RPis.
I do not use primitives in my rules, and since the problem started I do not copy all the files as a batch.
Whenever I make a change - it is only to one rules file, and I upload only that file.
I also read your recommendations here, some of which I read even before writing my rules.
I have no Thread::sleep in my rules - instead I usually use items with the expire binding. Sometimes I use createTimer.
I do not use executeCommandLine in my rules.
I do use sendHTttp*Request, to start/stop video recording and send email with video when my door is opened:
sendHttpGetRequest("http://192.168.1.99:8080/record?ot=2&oid=1)
Many of these are within a createTimer, in order to make sure the rule is not running for too long:
createTimer(now.plusSeconds(1), [ |
logInfo(âEntranceDoorOpenedâ, âBEGINNING: Start iSpy Recordingâ)
sendHttpGetRequest(âhttp://192.168.1.99:8080/record?ot=2&oid=1â) // Start iSpy Recording
logInfo(âEntranceDoorOpenedâ, âEND: Start iSpy Recordingâ)
])
Since I cannot find any solution to this new issue, I have started re-installing openhabian, hoping that this will solve my issue (maybe something was corrupted with the upgrade to 2.5 (and last nigh also downgrade back to OH2.4).
Are there additional things you recommend I should check in my rules?
I really try to be a âgood studentâ here, but I donât know what I am missing.