"WRITE ERROR on recording, disk full?"

[birdman]

This has started again for me, currently using LX3
It was better with ATV

Something from the logs below.

The first one shows an Interrupted System call, as seen previously (see #16 and #25). However, it is different in that it has happened quite soon after the recording started. You seem to have had two concurrent recordings on the go (consecutive episodes?) and the other one kept going, which at least seems to answer the question of whether the disk really was full.

The second excerpt is different. Not sure how it relates to anything else going on (why not upload the while log?). It looks like what I would expect to happen if the disk subsystem were to stop/be slow handling write requests? (although the EOVERFLOW itself appears to be a result of not reading sufficiently quickly from the tuner data).

Are you writing to a network share here, or a local/usb disk?
Do you have timeshift on and, if so, where is that writing?

 

[birdman]

Well, I can reproduce (forcibly) the "Interrupted System call" messages in a way that might explain the log in #16.
It is highly unlikely to explain the first part of #40.

What I did was to run (temporarily) enigma2 with a limit on the size of an output file of 40000 (which is 20MB on my box) and start a recording. The result was:

< 831.328208> [eDVBRecordFileThread] poll: aio_return returned failure: Interrupted system call
< 831.328680> [eFilePushThreadRecorder] WRITE ERROR, aborting thread: Interrupted system call
< 831.328967> [eDVBServiceRecord] record write error
< 831.329136> [eDVBServiceRecord] stop recording!
< 831.329989> [eDVBRecordFileThread] waiting for aio to complete
< 831.330578> [eDVBRecordFileThread] wait: aio_return returned failure: Interrupted system call
< 831.330726> [eDVBRecordFileThread] buffer usage histogram (20 buffers of 188 kB)
< 831.330958> [eDVBRecordFileThread] 0: 3
< 831.331360> [eDVBRecordFileThread] 1: 105
< 831.335321> [eFilePushThreadRecorder] stopping thread.
< 831.363607> [eFilePushThreadRecorder] THREAD STOP
< 831.368150> [eDVBTSTools] setSource loading streaminfo for /media/hdd/movie/20160103 1757 - BBC TWO - The Great History Quiz_ The Tudors.ts
< 831.369766> [eDVBServiceRecord] fixed up 1a84ca942 to bc92 (offset 0)
WRITE ERROR on recording, disk full?
RemovePopup, id = DiskFullMessage
AddPopup, id = DiskFullMessage
[SKIN] processing screen MessageBoxSimple:
[SKIN] processing screen MessageBoxSimple_summary:

One interesting(?) thing was that after this had happened the timer was still showing as active (recording) in the timer list.

Anyway, back to ymeknassi's problem. You'll need to logjn to the system, run "ps -ef" and look for the pid (second column) of the /usr/bin/enigma2 process (not the enigma2.sh one), then "cd /proc/<<pid>>" for that number followed by "cat limits". The second limit (Max file size) should be unlimited for both the Soft and Hard limit. Is it?

 

[ymeknassi]

Hi guys, I was quite busy the last days and didn't follow up on this. Anyway back from BlackHole where recording is working correctly; I installed now open vix 3.2 . Same issue is still happening. I also checked as per your recommendation and both limits are Unlimited

Well, I can reproduce (forcibly) the "Interrupted System call" messages in a way that might explain the log in #16.
It is highly unlikely to explain the first part of #40.

What I did was to run (temporarily) enigma2 with a limit on the size of an output file of 40000 (which is 20MB on my box) and start a recording. The result was:
One interesting(?) thing was that after this had happened the timer was still showing as active (recording) in the timer list.

Anyway, back to ymeknassi's problem. You'll need to logjn to the system, run "ps -ef" and look for the pid (second column) of the /usr/bin/enigma2 process (not the enigma2.sh one), then "cd /proc/<<pid>>" for that number followed by "cat limits". The second limit (Max file size) should be unlimited for both the Soft and Hard limit. Is it?

 

[Photovoltaic]

If you're running Samba on a Linux system..why not just ditch Samba and use NFS instead? I use an NFS Mounted 16TB Filesystem as my "Hard Drive Replacement" mount and record all content to it. (OpenViX, Xtrend ET10000)

 

[rassi]

The first one shows an Interrupted System call, as seen previously (see #16 and #25). However, it is different in that it has happened quite soon after the recording started. You seem to have had two concurrent recordings on the go (consecutive episodes?) and the other one kept going, which at least seems to answer the question of whether the disk really was full.

The second excerpt is different. Not sure how it relates to anything else going on (why not upload the while log?). It looks like what I would expect to happen if the disk subsystem were to stop/be slow handling write requests? (although the EOVERFLOW itself appears to be a result of not reading sufficiently quickly from the tuner data).

Are you writing to a network share here, or a local/usb disk?
Do you have timeshift on and, if so, where is that writing?

HDD (NFS mount) is not full, there is always something like 3-7TB free. It seem that it dont matter do i have just SD or SD+HD+HD recording going on
Attached log from today. There is one [RecordTimer] WRITE ERROR on recording, disk full?

zipped
View attachment Enigma2-2016-02-16_08-27-56.zip

If you're running Samba on a Linux system..why not just ditch Samba and use NFS instead?

It do not make any difference. Ok at least in my case. Using NFS.

 

[birdman]

zipped
View attachment 46821

This line might mean something (although it is almost 4 mins earlier, the I/O is asynchronous). Need to check whether the other logs show the same (EDIT - no they don't...):

< 27430.899961> [eMainloop::processOneEvent] unhandled POLLERR/HUP/NVAL for fd 90(16)

And 90 file-descriptors sounds like a lot of open files - but possibly not....

 

[rassi]

This line might mean something (although it is almost 4 mins earlier, the I/O is asynchronous). Need to check whether the other logs show the same (EDIT - no they don't...):

< 27430.899961> [eMainloop::processOneEvent] unhandled POLLERR/HUP/NVAL for fd 90(16)

And 90 file-descriptors sounds like a lot of open files - but possibly not....

For me that line wont say anything
I´ll try to see do I still have full logs from earlier issues, when posted only partials.

 

[ymeknassi]

I'm actuammu using a network hdd. Lacie cloudbox. It's very limited in term of flexibility: I cannot change filesystem or share type or anything. My only option is windows samba share. I'll never buy this brand again BTW. I'm an open source lover
I dont' have any simultaneous recordings or double processing running during the recording. Also, I tried blackhole is the only image where the issue doesn't occur. It happens is openvix, openpli and openatv exatcly the same way: at the same file size

Sent from my E5823 using Tapatalk

 

[rassi]

Now i´ve completely changed whole infra and issue still on.
NFS server changed, switches changed, Enigma2 box changed and still OpenViX/OpenPli variant causes Disk full error.
Vu+ DUO with Blackhole never had issues. Currently using LX3 with gbit. Newest logs attached.


View attachment Enigma2-2016-03-01_14-18-19.zip

 

[birdman]

Newest logs attached.


View attachment 47061

That one is different. It looks like the network writing was too slow and it ran out of local buffers to put the recording data in.

 

[rassi]

That one is different. It looks like the network writing was too slow and it ran out of local buffers to put the recording data in.

Thanks. Yes it seem that NFS performce is currently VERY poor on my gbit network from Xpeed LX3. Now i need to figure, why
It´s been MUCH faster, almost what you just can throw over the gbit ethernet.
No issues transfer from other HW to the same servers.

root@xpeedlx3:~# dd if=/dev/zero of=/media/tmp/testfile2 bs=512k count=1000
1000+0 records in
1000+0 records out
524288000 bytes (500.0MB) copied, 53.080058 seconds, 9.4MB/s

root@xpeedlx3:/# ethtool eth0
Settings for eth0:
        Supported ports: [ TP MII ]
        Supported link modes:   10baseT/Half 10baseT/Full
                                100baseT/Half 100baseT/Full
                                1000baseT/Half 1000baseT/Full
        Supported pause frame use: No
        Supports auto-negotiation: Yes
        Advertised link modes:  10baseT/Half 10baseT/Full
                                100baseT/Half 100baseT/Full
                                1000baseT/Full
        Advertised pause frame use: Symmetric
        Advertised auto-negotiation: Yes
        Link partner advertised link modes:  10baseT/Half 10baseT/Full
                                             100baseT/Half 100baseT/Full
                                             1000baseT/Half 1000baseT/Full
        Link partner advertised pause frame use: Symmetric Receive-only
        Link partner advertised auto-negotiation: Yes
        Speed: 1000Mb/s
        Duplex: Full
        Port: MII
        PHYAD: 4
        Transceiver: internal
        Auto-negotiation: on
        Supports Wake-on: ags
        Wake-on: d
        SecureOn password: 00:00:00:00:00:00
        Link detected: yes

Mount:

192.168.1.2:/mnt/tallenteet on /media/hdd type nfs (rw,relatime,vers=3,rsize=16384,wsize=16384,namlen=255,soft,nolock,proto=tcp,timeo=14,retrans=2,sec=sys,mountaddr=192.168.1.2,mountvers=3,mountproto=tcp,local_lock=all,addr=192.168.1.2)

fstab.
async is set, but it seem to not using it.. trying it to two different server, first ZFS RAIDZ and second RAID5. Both running Ubuntu.

192.168.1.2:/mnt/tallenteet     /media/hdd      nfs     _netdev,rw,nolock,tcp,async,nfsvers=3,rsize=16384,wsize=16384,timeo=14,soft     0 0
192.168.1.6:/mnt/video     /media/tmp      nfs     _netdev,rw,nolock,tcp,async,nfsvers=3,rsize=32768,wsize=32768,timeo=14,soft      0 0

exports are equal on both servers. except path.

/mnt/tallenteet 192.168.1.0/24(rw,no_root_squash,no_subtree_check,async)

 

[rassi]

Back on track. Bad connector on network. Now need to wait new "Write errors"

root@xpeedlx3:~# dd if=/dev/zero of=/media/hdd/testfile2231 bs=512k count=1000
1000+0 records in
1000+0 records out
524288000 bytes (500.0MB) copied, 10.616792 seconds, 47.1MB/s