[djmakinera]

True, during the closing JD2 (process javaw.exe uses much more RAM)
The application (process javaw.exe) for a long time before it closes probably save)

[djmakinera]

It simply fails to save the list because it is out of memory!

How much need to reserve physical RAM to run jdownloader with

[jd.controlling.linkcollector.LinkCollector$16(run)
CollectorList found: 10564/1157667

java.lang.OutOfMemoryError: Java heap space
at java.io.BufferedOutputStream.<init>(Unknown Source)
at jd.controlling.linkcollector.LinkCollector$23.run(LinkCollector.java:2315)
at jd.controlling.linkcollector.LinkCollector$23.run(LinkCollector.java:2249)
at org.appwork.utils.event.queue.QueueAction.start(QueueAction.java:202)
at org.appwork.utils.event.queue.Queue.startItem(Queue.java:491)
at org.appwork.utils.event.queue.Queue.runQueue(Queue.java:425)
at org.appwork.utils.event.queue.QueueThread.run(QueueThread.java:64)

[Jiaz]

Thanks for the logfile. Now I know what you mean, Please update next core update in a few minutes

[djmakinera]

Jiaz: Reduced memory usage during linklist saving, please check again with next core update

New update - causes the loss of links (!!!)
1 mln - 1,4 mln links
Update causes damage to the archives!

Previous Size "LinkCollector": 657 MB

Now
1,9 MB !!! zip.backup
9 MB linkcollector.zip

Previously in the previous version JD2 (example date 30-10.16) NEVER damage archive even if forcibly cut off power supply (so better!)
Old Archive rename larger number, but in the future again and again and again cause damage. A very serious bug!

[djmakinera]

Previous Version: (...) (...) - 30.10.2016 work
In previous logs is not has performed this entry, now find all the time

New:
This causes damage to the removal and proper archives (even old archives !!!)
All the time the loss of links

------------------------Thread: 95:Log.L.log-----------------------
--ID:95TS:1478004548042-01.11.16 13:49:08 -  [org.appwork.shutdown.ShutdownController$3(run)] -> Exit Now: Code: 0
------------------------Thread: 21:Log.L.log-----------------------
--ID:21TS:1478004548421-01.11.16 13:49:08 -  [] -> java.net.SocketException: socket closed
	at java.net.TwoStacksPlainSocketImpl.socketAccept(Native Method)
	at java.net.AbstractPlainSocketImpl.accept(Unknown Source)
	at java.net.PlainSocketImpl.accept(Unknown Source)
	at java.net.ServerSocket.implAccept(Unknown Source)
	at java.net.ServerSocket.accept(Unknown Source)
	at org.appwork.utils.singleapp.SingleAppInstance$1.run(SingleAppInstance.java:364)
	at java.lang.Thread.run(Unknown Source)

[editestowy]

Not confirmed - your screenshot doesn't prove anything.
How do we know that you didn't delete them by yourself?

Here are mine list - before (21:47) and after (21:51) latest core update:


Nothing was lost after updating, filesize was updated after I've moved 2 packages into Downloads.

[djmakinera]

editestowy -

1. Misread my previous post!!!
Linkcollector!!!
(NOT(!) DownloadList
2. You tested with 1,000,000 - 1,400,000 Link??? 600 - 700 MB ! Certainly not! So I do not understand your answer!

Jiaz had to update the core (was supposed to reduce the use of creating a backup that caused damage to the archive eg. After a long terminate process javaw.exe (more than 60 seconds TimeOut! Socket Exception Error)

I do not like the new update
https://board.jdownloader.org/showpo...5&postcount=19
https://board.jdownloader.org/showpo...6&postcount=20
because it bring catastrophic consequences.
I'm going back to the old version. I'm sorry! I'm telling the truth.

Reproduce
a) 1+ mln Links (600-700 MB)
b) Restart JD2
WARNING: Unexpected end of archive 0000 (broken) (1 MB)

[djmakinera]

New update - unexpected bad results
Why after a restart JD2 damages archive (Practically all the time)
Note (!) Also removes Xold archives
Closing the application will take several minutes.
Perhaps it may work for 100,000 - 200,000
but it does not work correctly for 1,400,000 K links (!) CollectorLinks

[Jiaz]

Please provide complete logfile. Nothing changed and that you have links loss is the same reason for your other bugreport about those empty/zero size files. It is the exact same cause.
The Stracktrace you've provided has nothing to do with this

[Jiaz]

Originally Posted by djmakinera

This causes damage to the removal and proper archives (even old archives !!!)

Please stop telling lies. Only difference is that instead of writing everything to memory first and then one big write to disk (which caused the high memory usage for you) it is now written with smaller buffer. So in fact it is the same issue you suffer all the time that the shutdown has not enough time to write list and instead of having a zero size file (before) you now have a half finished file.

[djmakinera]

f I'm lying, please withdraw this update. I repeat once again, please remove this update. I do not need an update, which causes damage to (the loss of links) Thanks.
Previously, no damage archive LinkCollector (only bakup) 0byte.
That's all from my side!
I do not have the strength to prove everything every time (this is my personal opinions / testing)
Sorry for wszytsko. I'm just experiencing problems so he writes.
You do not believe me. Excuse me :(

[djmakinera]

Please stop telling lies.

If I lie, why do you have over 1000K+ links in Linkgrabber after the RE-start jd2 I have 2000 links.
Note very successfully re-START JD2 (not to be confused with the launch (run JD2) !!!) It is unfair suspicion if someone does not have a million links and test :(

[djmakinera]

Originally Posted by Jiaz

Please stop telling lies.

I'm not a liar. It's too big a word. :-(

[djmakinera]

02.11.16 10.50.48 <--> 02.11.16 10.51.03 jdlog://3681881887641/

[djmakinera]

New Update
I set -Xmx11 GB is too little? for 1 million links?
Why damages zip archive XXX.zip & XXXzip.bakup and lose links after restart.
Please answer?


Old Version -Xmx6G or 7G - support 1+ mln links (new version - NOT SUPPORT) corrupt archiwe when restart JD2

[djmakinera]

How to: always autorecover links after a Run or restart JD2?
I mean more than 10,000 packages and millions of links?

[Jiaz]

This is still the same error as before.
Before you had zero byte Files and now you can see that JDownloader has not enough time to save the complete list.

Do you close JD normally or shutdown Computer?

[Jiaz]

Originally Posted by djmakinera

Previously, no damage archive LinkCollector (only bakup) 0byte.

Are you kidding me. I will not change it back. Because before it also did not work for you. a 0byte file is the same (broken) as a half written fie (broken).

The question is, why is the file broken/not finished. So do you close JDownloader normally or kill it (eg on computer shutdown)?

[Jiaz]

Originally Posted by djmakinera

New Update
I set -Xmx11 GB is too little? for 1 million links?
Why damages zip archive XXX.zip & XXXzip.bakup and lose links after restart.
Please answer?


Old Version -Xmx6G or 7G - support 1+ mln links (new version - NOT SUPPORT) corrupt archiwe when restart JD2

This has nothing to do with memory! It is the exact issue I've told you many times again.
It is the same cause for 0 byte files. Now JDownloader is able to start writing the files but because of unknown reason JDownloader is killed before saving the list.

[Jiaz]

Before the update:
Write 500 mbyte to memory and then begin to write to disk
->result: 0 byte file written, broken list

Now:
Directly write to disk with smaller buffer (1 Mbyte max)
-> result: file size depends how long JDownloader before it gets killed, broken list

HOW do you do it?
Please the exact steps!


This problem only happens for you and we need to find out why.
I've just tested with 500k links and 10k packages. all fine