16 replies to this topic

[Hakon]

Having problem with recordings to FreeNas server after upgrade to PLi 3.0.
The recordings freeze in a few seconds , so playback of these recordings is not fun.
Same problem with SD as HD recordings.
If i playback "old" recordings that was done with PLi 2.0 its no problem.



Had same issue a few years ago when i upgrade to 2.0 (i think it was 2.0), i did get help here with edit something but i can't remeber how to do it.
Tried to find the thread but might be to old.

Vu+Duo
FreeNas server
Wired gigabit lan

[WanWizard]

Are you using CIFS or NFS? How did you mount the NAS?

[Hakon]

Mounting NFS

Mountsyntax in fstab
192.168.0.xxx:mnt/nas5 /media/hdd nfs rsize=8192,wsize=8192,rw,nolock 1 0

[catastrofus]

Try using udp instead of tcp (which is the default)

[korsan]

Which box do you have?
I had this same problem too on Openpli3.0 VuDuo + nas recordings. I solved this by adding 64mb swap on usb. With 2.1 I never had this prob.

[hemertje]

Which box do you have?
I had this same problem too on Openpli3.0 VuDuo + nas recordings. I solved this by adding 64mb swap on usb. With 2.1 I never had this prob.

Are you using CIFS or NFS? How did you mount the NAS?

Using NFS use "wsize=4096,rsize=4096"

[Hakon]

I found my old topic _http://openpli.org/forums/topic/16483-vuduo-recordings-freeze-after-update/

When i use this commands it does not freeze

Are the freezes in the recordings (that is, if you play back it will freeze at exactly the same spot every time) ?

If so, try the following commands in telnet on the recording box (with OpenPLi 2.0), and see if that makes the problems go away:

echo 1 > /proc/sys/vm/dirty_background_ratio
echo 60 > /proc/sys/vm/dirty_ratio

I will test the "Using NFS use "wsize=4096,rsize=4096" " to see if it's helps.

Thanks Hakon

Edited by Hakon, 17 February 2013 - 11:11.

[Hakon]

Which box do you have?
I had this same problem too on Openpli3.0 VuDuo + nas recordings. I solved this by adding 64mb swap on usb. With 2.1 I never had this prob.

I have Vu+Duo x 2 , same problem on both boxes.

As i understand it's the memory that's running out.
Im have my hope that the Duo2 will soon be on the market, then no problem with memory ...i believe.

Edited by Hakon, 17 February 2013 - 11:18.

[Hakon]

Oops , i now se that i have started this topic in wrong forum.
It should be in the (EN) Enduser support

Is there a Moderator who can move this topic to the correct forum ?

Thanks Hakon

[WanWizard]

Topic moved.

[atilaks]

echo 1 > /proc/sys/vm/dirty_background_ratio
echo 60 > /proc/sys/vm/dirty_ratio

omg, omg... it finally solved my problem with NFS... thank you!!!

[Hakon]

I have made a test with a swap .
The first "free" command result is without recordings.
Next "free" command is with 1 HD channel recording, next with 2 HD channels recordings , and so on to 4 channels recordings.
The swap are more and more used.
Can some explain whats going on ?

The testrecordings was made for 9 minutes, and still freeze. In this test no echo 1 > /proc/sys/vm/dirty_background_ratio echo 60 > /proc/sys/vm/dirty_ratio.


No recordings
total used free shared buffers
Mem: 138164 135588 2576 0 1096
-/+ buffers: 134492 3672
Swap: 1048572 0 1048572

One HD channel recording
root@vuduo:~# free
total used free shared buffers
Mem: 138164 135156 3008 0 340
-/+ buffers: 134816 3348
Swap: 1048572 632 1047940

Two HD channel recording
root@vuduo:~# free
total used free shared buffers
Mem: 138164 134984 3180 0 112
-/+ buffers: 134872 3292
Swap: 1048572 2676 1045896

Three channel recording
root@vuduo:~# free
total used free shared buffers
Mem: 138164 135280 2884 0 112
-/+ buffers: 135168 2996
Swap: 1048572 2664 1045908

Four channel recording
root@vuduo:~# free
total used free shared buffers
Mem: 138164 135120 3044 0 72
-/+ buffers: 135048 3116
Swap: 1048572 8296 1040276

Edited by Hakon, 17 February 2013 - 14:07.

[Hakon]

And a test with the command
echo 1 > /proc/sys/vm/dirty_background_ratio
echo 60 > /proc/sys/vm/dirty_ratio.

No freeze .

Attached Files

•  Skärmbild från 2013-02-17 14:22:25.png   392.66KB   32 downloads

[WanWizard]

free doesn't tell you anything, so it's pretty pointless looking at it.

As mentioned in several other threads, the issue with NFS is that the new 3.x kernel by default allocates buffers that are too big, and which must be continuous memory blocks. The box doesn't have many of those, and over time this gets worse, so you get these issues due to lack of buffer space. This is not solved by adding swap, and changing the kernel parameters to flush more often is a workaround.

The only correct solution is to make sure it allocates smaller buffers, which you do by specifying rsize and wsize on the mount. What the correct values are differ from user to user, so you'll have to experiment a bit. 4K or 8K buffers seem to work fine for most people.

[Hakon]

Ok , thanks for your statement.
I will make a test with the 4K rsize and wsize.

[Hakon]

Have made a lot of recordings with the 4K rsize and wsize, but same problem.
The freeze are now longer when it appears.
So this was not the solution for my system.

The only thing that works is the

echo 1 > /proc/sys/vm/dirty_background_ratio
echo 60 > /proc/sys/vm/dirty_ratio

[BuGless]

Have made a lot of recordings with the 4K rsize and wsize, but same problem.
The freeze are now longer when it appears.

What does "mount" say about the current active mount options in use for your NFS partition?