• 1
  • 3
  • 4
  • 5(current)
  • 6
  • 7
Rasberry PI lag - can we fix it?
#61
Haven't tested with the hub yet, still using Yatse but it just restarted XBMC when I went into Video - Files - Add Videos and went to Browse for a source. When it rebooted it said "Remote communication failed" which is the message I've seen a lot before (I think that's exactly what it said, I keep forgetting by the time I type these messages!)

Just before this I'd gone to File Manager - Add Source and added a shortcut to my NFS server on 192.168.1.64 to make it easier to access from Video, so I don't know if that's connected.

Before that I'd installed xTV-SAF skin, Confluence Modified, Youtube and a few radio plugins.

log from before the restart http://pastebin.com/3ndfraSA (I had to edit it down as it was about 9MB!)

and after http://pastebin.com/ZTSk4rYD

and dmesg from after http://pastebin.com/Pb3GdVwY (no out of memory messages that I can see)
Reply
#62
In the after log the only failed messages I see are

20:27:55 T:3069612032 ERROR: JSONRPC Server: Failed to connect to sdpd
20:27:55 T:3069612032 ERROR: JSONRPC Server: Failed to bind serversocket
20:27:55 T:3069612032 ERROR: AIRPLAY Server: Failed to bind serversocket

20:27:55 T:3069612032 ERROR: Failed to start AirTunes Server
21:11:59 T:2859836512 ERROR: NFS: Failed to mount nfs share: (nfs_mount_async failed)

I don't know if one of those generated the "Remote communication failed" error though, or if they matter.

I'm going to try removing the XBMC NFS share and making a OS NFS mount instead using automount.sh and

mount -t nfs 192.168.1.64:/Media -o nfsvers=3,rw,intr,noatime,rsize=32768,wsize=32678,nolock,async,proto=udp /storage/NFS-Media

and see if that helps.
Reply
#63
(2013-07-22, 19:56)popcornmix Wrote:
(2013-07-22, 19:29)doveman2 Wrote: Does the fact that it rebooted before when plugging in the power to the hub (with no other leads/devices connected) suggest that it's not possible to power the Pi and hub from one ubec (although I know others have done this without any problems)? It did also reboot once when I was plugging/unplugging a USB stick into the hub and obviously that's not good as we need to be able to plug in a stick to play media without it rebooting.
Both those statements sound concerning. Is the hub back-powering? (i.e. if you are connected to HUB, but remove Pi's power lead, does it continue to run?)
http://www.raspberrypi.org/phpBB3/viewto...=29&t=8261

Thinking about it, as it reboots when connecting the power to the hub with nothing else connected to it, it can't be a back-powering problem.

I just tested and this happened again. Connecting the usb lead after didn't cause it to reboot and pulling the power from the GPIO caused it to shut off.

Unplugging/plugging the remote receiver from either of the two external ports on the hub didn't cause it to reboot either.
Reply
#64
Hmm, the OS NFS mounts aren't working very well. I can't play anything from it and it just shows the busy icon for minutes before dumping me back to the library.

After I tried a few movies and backed out to the parent folder, then it froze for a while again and then started playing something but froze after a few frames and I had to reboot via SSH.

I tried the same movies via SMB after rebooting and they worked fine. I was also able to play them previously when using the XBMC NFS mounts, so maybe there's something about this command and my Hanewin settings that isn't compatible.

Log here http://pastebin.com/M2uKZXmu
Reply
#65
You are not getting reliable data streamed over network.
Code:
22:35:05 T:2843890784   ERROR: COMXImage::ReadFile /storage/NFS-Media/Movies/Movies to Watch/Drama/A Single Man (2009).tbn GetCodingType=0x7fffffff
22:35:24 T:2987029600 WARNING: JpegIO: Error 55: Not a JPEG file: starts with 0x%02x 0x%02x
22:35:24 T:2843890784   ERROR: Texture manager unable to load image from memory
22:36:15 T:2987029600   ERROR: ffmpeg[B20A7460]: [ac3] frame sync error
22:38:17 T:2987029600   ERROR: ffmpeg[B20A7460]: [mp3] Header missing

I use Hanewin with no issues. I set it up using settings like:
http://www.networkedmediatank.com/wiki/i...NFS_server

Any errors in dmesg log?
Reply
#66
Thanks, I got it working by removing proto=udp from the mount command but I'll try changing the Hanewin settings and add it back.

I currently have Max NFS Transfer Size set to 8192, NFS v2 as well as v3 enabled, and don't have Allow Mount of Remote Devices enabled or ANSI (or UTF-8) ticked, so maybe one of those affects UDP but not TCP.

EDIT: Nope, none of that helped. It took ages waiting, then started the movie and immediately froze, pegging the CPU at 91% for a while (now at 18%).

There's some "Server not responding" errors at the end of dmesg http://pastebin.com/70L60bRm

Anyway, it's not important and I can just use tcp for now. Sorry for taking this thread off-topic, I only posted about this problem here because of the previous restarts that might have been connected to the XBMC NFS mounts but the OS mounts seem OK so far Smile
Reply
#67
I've found plugging the remote receiver into either port on the Pi causes it to reboot, whereas plugging it into the hub doesn't, which I suppose indicates that the hub's power supply is somewhat separate, although I guess when I plug the power into the hub it causes a fluctuation of sorts on the ubec output, which causes the Pi to reboot.

I'm not sure why the Pi would reboot when plugging a USB device into it as I'd think it was intended that people would plug sticks in at least but I'm not really bothered about that as I don't intend to be doing this and my focus is on getting the hub and remote working.
Reply
#68
Actually, sometimes when trying to shutdown, the screen doesn't even blank and I only know that XBMC isn't still running because the debug OSD stops updating.

I initiated the shutdown at 14:27:43 (13:27:43 according to the log) and it's ten minutes later and it's still like that, although the log stopped after 10s when XBMC presumably was terminated).

http://pastebin.com/

Fudge, my previous post seems to have disappeared!

Right let's see if I can remember what I posted.

It seems to be pretty stable now, even with the hub and remote connected, albeit sometimes lagging/ignoring keypresses and occasionally going into a phantom keypress loop (although that seems much less likely to happen now). I don't understand why the remote would lock up sending phantom keypresses whenever I try it in the Pi directly, as I'd think that should be less complicated but hey.

I actually swapped the hub out as I wanted to try one (same model) I'd disconnected the LEDs from, in the hope that it would generate less heat and thus make my box less warm. It turned out that it doesn't help and the heat is being generated by the current draw by the remote receiver (without that, or anything else, connected it doesn't get warm) but maybe the other hub was just dodgy. I'll try swapping it back in to test.

I've got iPlayer, ITV Player, Youtube, Radio, ListenLiveEu, ShoutCast2 working reasonably. Sometimes the radio plugins lockup the GUI for as much as a minute when trying to start a station with a non-working link and Shoutcast2 seems to be a lot slower than Radio, taking about 30s to switch stations but other than that they're OK.

When booting I see some messages but they're probably not critical.

sbin/init: source: line 49: can't open '/var/config/settings.conf.
rpcbind: cannot get local address for udp. Servname not supported for ai_socktype
same message for tcp then udp6 and tcp6 several times.
garbage after 'name' token in line 378 ignored
invalid code found for Xbox: KEY_PAUSE

I don't seem to be getting the rainbow splash most of the time but just linux boot text, or sometimes I get both.

Also, sometimes I can't seem to shutdown properly now and although the screen goes blank, all the lights stay on, with Link flickering. I don't know if this could be caused by running from SD and using iotimeout maybe, or by using autostart.sh to make an OS-level NFS mount at boot, like this (comes in handy to pause at boot to note any error messages, as I can just disconnect the network and then as it can't find it, it pause the boot).

Code:
#!/bin/sh

# Wait for the network to come up...
while [ -z "$(connmanctl state | grep State | grep -E "ready|online")" ]; do sleep 0.25; done; logger -t "$(basename $0)" "** Network is up **"

[ ! -d /storage/NFS-Media ] && mkdir /storage/NFS-Media

mount -t nfs 192.168.1.64:/Media -o nfsvers=3,rw,intr,noatime,rsize=32768,wsize=32678,nolock,async,proto=udp /storage/NFS-Media

Other than the power-off problem though it seems stable enough, so I should try moving Storage back to the USB stick next and then try overclocking again, although I guess we should try and fix the power-off problem first (unless it is caused by running from SD and using iotimeout, in which case it might not be a problem when using the USB stick for Storage, I certainly never experienced it when I was using the USB stick before resetting .xbmc)
Reply
#69
(2013-07-23, 15:40)doveman2 Wrote: I don't know if this could be caused by running from SD and using iotimeout maybe, or by using autostart.sh to make an OS-level NFS mount at boot,

rbej stopped supporting the iotimeout (anti-sd card corruption check) functionality some time ago.

Can't think what is causing the shutdown problem, shouldn't be the OS mounts.
Texture Cache Maintenance Utility: Preload your texture cache for optimal UI performance. Remotely manage media libraries. Purge unused artwork to free up space. Find missing media. Configurable QA check to highlight metadata issues. Aid in diagnosis of library and cache related problems.
Reply
#70
Oh really. Thanks, I'd better remove that then.

I'll have to just see if the shutdown problem still happens at all when using the USB stick for storage as I was before, although I wouldn't think that would make a difference.
Reply
#71
Well I tested with Frodo 12.1 on the SD, than rbej's OE Gotham on the SD, then Frodo using the USB stick for storage, then Gotham using the USB stick for storage and they all seem rock solid, even with the hub and remote receiver connected, which seems to be working much better now without lagging and only the very occasional phantom keypress loop.

Then I put the original hub (with the LEDs still intact back in) and that seemed OK still although I didn't test it for ages. Then I enabled the previous overclock

arm_freq=900
core_freq=333
sdram_freq=450

and with that, when I went to go into File Manger it froze up (debug OSD stopped updating) and I can't even access it via SSH or SCP, which I normally can when it freezes.

I'll put the LED-less hub back in and see if that's any better but it's looking like this board just doesn't like being overclocked. I've got another two boards though, so I'll try with those as it really does need to be overclocked to run smooth enough to be acceptable.

With the LED-less hub it seemed more stable (didn't lock up going into File Manager and played a Movie over my OS-level NFS share) but then I tried to play an iPlayer TV stream and first try it played about one second and stopped and the second try it rebooted and said "Remote communication server - Failed to Start".
Reply
#72
Well with my second board, non-overclocked and using the unlit-hub, it froze when I pressed Fullscreen to switch from Settings back to the movie (playing over the OS-level NFS share)

It hasn't completely frozen and the debug OSD is still updating but it's completely unresponsive and top shows it's only using less than 10%. I just mashed some buttons and they've been logged, or some of them at least (at 01:37:46) but there was no response from XBMC. http://pastebin.com/FgudVuus

I can't ls in /storage in a SSH shell, nor can I navigate in WinSCP anymore and now that can't re-connect either. The movie is still showing (frozen) on my TV though!

EDIT: It occurred to me that perhaps I NEED to use over_voltage=2 with that overclock. I wasn't because of what it says in the config "try to use Medium mode without "over_voltage=2" if experiencing corruption with Turbo Mode, so I assumed that Medium settings should work without overvoltage but I suppose if I'm overclocking it makes sense that it would need more voltage.

Anyway, I'm back on the first board now using Medium with over_voltage=2 and it seems fine so far Smile

The second board looks like it might be a bit of a dud though if it can't even work reliably without any overclock Sad
Reply
#73
Well I'm not sure this first board is fine yet. I was watching a mpeg2/ts recording over SMB and when I stopped it I just had a black screen for about a minute (03:08:44 -03:09:49), with the debug OSB frozen.

Couldn't paste the whole log, so I chopped the middle bit out, where I was using iPlayer http://pastebin.com/ABSMp5Fm
Reply
#74
Try Xbian - usually much faster than Open Elec
Reply
#75
I tried

arm_freq=1000
core_freq=500
sdram_freq=600
over_voltage=6

but that hung after the linux boot text and before the rainbow splash

arm_freq=1000
core_freq=500
sdram_freq=500
over_voltage=6

went further but it showed a lot more linux boot text than usual and ultimately hung after mounting the USB stick

arm_freq=950
core_freq=450
sdram_freq=450
over_voltage=6

works OK, although I did get a phantom keypress loop almost straight away (in the shutdown menu).

I've got initial_turbo=30 set, which if I recall correctly is meant to prevent the overclock from kicking in for a while during boot but I might have misunderstood.

After booting and once the XBMC Home screen appears, the debug OSD shows the CPU is high (85%+) and the used RAM is changing rapidly for 27s before it stops and settles down to about 45%.
Reply
  • 1
  • 3
  • 4
  • 5(current)
  • 6
  • 7

Logout Mark Read Team Forum Stats Members Help
Rasberry PI lag - can we fix it?2