Network buffering problem after upgrading to 8.95.3

I mentioned my problem in here, but then though to start a new topic.

So I have this TV Box based on S905 chip. It has 1GB of RAM and 8GB storage.
Wifi does not work because of some missing driver, but wired network works just fine - at least on 8.95.2

The problem is when I upgrade to 8.95.3 (or take one of the recent test builds) I can’t play files from network anymore. By network I mean Windows Network (samba) shares. On most files (mostly HD) it just keeps re-buffering every minute or so and the buffering takes quite long each time.

When I downgrade to 8.95.2 all comes back to normal and I can again enjoy a normal network playback.

Any ideas?

Do you have buffering for all network streams in the advancedsettings.xml?

Normally I did not have advancedsettings.xml

But then trying to solve the problem I was experimenting with different settings for the cache.

But none of them helped :frowning:

Sounds like a PITA. You got the buffermode 1 option on?

Can you see what sort of throughput you’re getting over the network?



Now, I measure the speed of copying a file from a samba share to local /tmp folder, using Kodi’s built in File Manager.

On 8.95.2 I get a speed of almost 1000 KB/s

On 8.95.3 I get less than 400 KB/s.

What’s interesting, when I use Windows Explorer to copy the tar file to the Upgrade folder - both the versions seem to be achieving similar speed: slightly more than 1.00 MB/s

FWIW, I tried with a different Ethernet adapter (external, connected via USB) and the results are pretty much the same.

Using Kodi’s File Manager, to read (copy) a file from a network server (samba share):

  • On 8.95.2 I get a speed of about 1000 KB/s
  • On 8.95.3 I get a speed of less than 400 KB/s

BTW, using Kodi running on Raspberry Pi 3, I get between 4 and 6 MB/s
So the network adapter of my TV Box is quite slow, in general.
But on 8.95.3 it is just too slow.

Anyone knows why?

ethtool returns the same info on both versions:

CoreELEC:~/.update # ethtool eth0
Settings for eth0:
        Supported ports: [ TP AUI BNC MII FIBRE ]
        Supported link modes:   10baseT/Half 10baseT/Full
                                100baseT/Half 100baseT/Full
        Supported pause frame use: No
        Supports auto-negotiation: Yes
        Supported FEC modes: Not reported
        Advertised link modes:  10baseT/Half 10baseT/Full
                                100baseT/Half 100baseT/Full
        Advertised pause frame use: No
        Advertised auto-negotiation: Yes
        Advertised FEC modes: Not reported
        Speed: 100Mb/s
        Duplex: Full
        Port: MII
        PHYAD: 8
        Transceiver: external
        Auto-negotiation: on
        Supports Wake-on: ug
        Wake-on: d
        Current message level: 0x0000003d (61)
                               drv link timer ifdown ifup
        Link detected: yes

I have a similar problem since 8.95.3 I have another thread open for. I was looking at the release notes and saw the blu-ray playback over Samba improvement and got me thinking. Need to check the git to see if it’s an isolated commit so I can determine what’s changed.

EDIT: This appears to be it:
EDIT: I think I will force SMBv3 and see what happens

Changed to 32K chunks from whatever was default.

I just realized this probably won’t help me (but may help you) as I stream via Emby over HTTP

@koler @flatline69
What SMB version are your servers running that have this problem?

openMediaVault 4 - I have it set to SMBv2/v3.

EDIT: I should note this doesn’t directly affect me as I stream via Emby (addon mode) which is over HTTP. I just noticed since 8.95.3 that after the box sits for a while, it can take up to 3 minutes for a title to start playing with a spinner just spinning. It’s fine directly after rebooting, almost instaneous.

I use Windows 10 computer as the server.

Try use NFS shares instead SAMBA and set advancedsettings.xml

I don’t have any problems and im using only 100mb lan box side.

Samba is fine. I ended up rebuilding the box (reflashed CE on it) and everything was fine from there on in. It’s been upgraded since the beginning, a refresh was clearly needed :slight_smile: