memory leak

[hellAT]

Hello,

since a few days Virtualbox is causing a memory leak on my system. I have got Ubuntu 18.04.3 and now, after updating the server, VB 6.1.2r135662 - but no change.

There are running two Win10 VMs on the host.

I don't know what happend, but now the memory is consumend after starting the VMs within 20-30 mins, what ends in unuseable VMs.


Any idea, where i can have look to find the issue?

regards
Karsten

[scottgus1]

Please zip and post a guest vbox.log for each of these guests, using the forum's Upload Attachment tab.

[hellAT]

okay, thnaks.

i've done so. See attachment

[scottgus1]

The idea I had doesn't seem to hold water, sorry. Your guests are 8 & 4.9 GB RAM, and the host is 32GB RAM, so there's plenty of room. (I had thought that the guests might want more than the host had, and the lazy memory allocation Virtualbox uses would allow the guests to start but then fill up the host.)

One other suggestion, if you have a spare hard drive lying about, try a fresh host OS on it, install Virtualbox, try the same two guests, and see what happens. Memory leaks don't get reported a lot, it seems. So maybe there's a glitch in your host OS?

Otherwise, I'm going to have to let the Linux gurus have at this one.

[hellAT]

Thanks for looking at this issue. Unfortunatly i have no spare harddisk at the moment with the right size.

So if there is somewhere a linux / vbox guru who might have a look on this...?

[fth0]

a memory leak on my system

If I understand you correctly, the memory of your Linux server is getting consumed. Can you elaborate a bit about that? How and what did you notice?

To get some more information on the Linux server's memory usage, for example you could try the following: Start the VMs, open a shell on your Linux server, execute top, press (capital) M to sort by memory usage and (capital) H to show threads instead of processes. Take screenshots of the top output after 5, 10 and 15 minutes, and shutdown the VMs directly afterwards. Post the VBox.log files and the screenshots, altogether in a zip file.

[hellAT]

I notice this at first by the users, complaining about the VMs, which are unuseable. Then i checked with "free" command, and recognized that only a few hundred Mbs are free.

I have collected the infos you mentioned, added some "free -m" output. I tried to access the VMs by GUI after about 25 mins - what did not work. But the Windows VMs have a SSH Deamon running, i could access by ssh, what was still working, and shutdown the VMs like you said.

[Martin]

There is no memory leak. Linux is using the free memory to cache disk access.
You need to look at the value for used memory (belegt) which doesn't change.
The amount of memory not yet used applications is the sum of free (frei) and buff/cache (Puffer/Cache), this is shown in the last colum as available (verfügbar).

[hellAT]

okay... but previously more memory was avail - and the VMs are working well.

What will help to make the stuff running again?

[fth0]

First of all, I come to the same conclusion as Martin that there is no memory leak on the host side. Additionally, the information provided by the top output doesn't indicate other host problems (e.g. CPU load, no waiting for disk I/O).

I notice this at first by the users, complaining about the VMs, which are unuseable. Then i checked with "free" command, and recognized that only a few hundred Mbs are free.

Since the suspected memory leak has turned out to be wrong (at least on the host side), I suggest to ask the users what unusable means ...