214 replies to this topic

[cmatte]

Cheers Erik for your insights!

In this case, there were no worries whatsoever on the RAM usage, however this seemed to be the driver of the box soft-bricking, with enigma2 using up to 1003m of RAM. A swapfile was added to support the eventual large usage of epg and picons, however even with 2GB on a fast usb drive, this was ultimately always making the device unusable and not even booting after a shutdown or reboot (black screen), solved with the workaround above.

 

After a lot of troubleshooting and testing, following the precious advice, a lot of improvement is experienced as for example enigma2 is now at 366m, thanks to the removal of CrossEPG jobs and the use of EPGImport, instead, still together with EPGRefresh. If this turns out to be the culprit and the usage stays under control, great! We'll know after a bit more testing, as there is only a 10 hrs uptime now and only a single cycle of EPGImport and EPGRefresh. If that was not the cause, there is still some unexpected memory usage growing until bricking the device to identify and troubleshoot 

[WanWizard]

It seems to be a fact that Enigma memory usage slowly increases for some reason.

 

Yesterday I reported 337m in use, now it has risen to 353m. Still, at this rate I have to leave it on for a year or so to get to your kind of numbers.

[mymon]

Ive upgraded from v6 to 7.1 and during recording can not list /hdd/move/

if do it via remote the spinner goes forever

if do it via cli it seems to be "freezing" as well

Has anyone seen this behaviour? I can roll back and do clean install but need confirmation its just my box

Thanks

[WanWizard]

No.

 

What is "/hdd/move"?

 

OpenPLi doesn't use "/hdd", but "/media/hdd", and only if you have a local harddisk.

[mymon]

correct.. /media/hdd/movie/

seems to be "locked" during recording

 

Note: sometimes when my PC/NAS is sleeping I have some mounted drives unreachable/down

but it wasn't an issue before

Edited by mymon, 7 October 2019 - 15:37.

[WanWizard]

Is that a local disk? Or a mount? And if so, what type of mount? With what mount options?

 

Obviously, if your PC or NAS is sleeping, the drives are down and the mount is unreachable. In which case I/O will hang, waiting for a timeout.

[MiLo]

EPG is a big suspect, it's stored in RAM in a very finicky format that will create memory leaks if anything goes wrong - and epg.dat will store and preserve any leaks. On the other hand, that should be obvious to see, if you reboot enigma and look at the epg.dat, it should be growing all the time too.

I've had boxes with 3 months and more uptime, so normal prolonged usage should not yield memory problems.

[mymon]

I had solo4k with 3month upload as well and no issues in 6.x

since upgraded to 7.1 the box is slow as its never been before

 

in relation to my local drive it has 50% free space

the ../movie folder has up to 40 files in it but also folders which are mounted drives like NAS / PC / etc

when Im not recording it lists the files but once recording goes on the spinner comes up and goes forever

 

the epg is 11M - can't tell what it was before as never needed to look into it

-rw-r--r--    1 root     root       11.0M Oct  7 05:11 epg.dat

If this is not a know issues then I guess I need to go for clean install

[WanWizard]

You don't have LCD4Linux installed by accident?

[Pike_Bishop]

Hi,

 

 

Note: sometimes when my PC/NAS is sleeping I have some mounted drives unreachable/down

but it wasn't an issue before

 

i have a similar problem here with my ultimo 4k and openpli-7.2rc but i had this problem in openpli-7.1 too

in fact i have a internal hdd and also a external usb-3.0 hdd, and i have a hdd on a router (fritzbox) for that i have a samba share with autofs and i have thereto a symlink

in the internal hdd (/media/hdd/movie) - that all works without problems, i have testet this the last two weeks.

 

but a have also a synology diskstation (nas) and for that a samba share too in the /etc/auto.net with autofs with a symlink in /media/hdd/movie for it

and with that i have the problem if i go in the movielist with movieselection that the movielist is open after till 2 minutes and i can see the spinner for this 2 minutes

, and also if i want stop a playback (not from the synology diskstation but from the internal hdd) it's once in a while the same problem till i can close the movielist

(till 2 minutes spinner)

that problems are not always (but often) if i want to go in the movielist or stop a playback and close the movielist.

 

i have testet in the last two weeks without all shares - no problem, all ok.

and then i have testet only with a share to another e2box and later with the share (hdd on my router) too and no problems, all ok

 

but if ia activate the share to my synology diskstation with symlink for it in the (internal hdd) /media/hdd/movie the problem is coming again.

i have also testet with a nfs share instead a samba share for the synology diskstation but it's the same problem.

 

so i know now that only the synology diskstation caused this problem, but only in pli image in openatv i didn't have that problem.

 

 

regards

Pike

Edited by Pike_Bishop, 7 October 2019 - 22:11.

[mymon]

thanks for the update.. I haven't spent much time digging but it does seem to be share/samba/cifs related

I just rolled back to 6.2 and box is flying it!

[Frenske]

I experience an equal issue, huge delay when opening the movielist on my remote box which is mounted to the harddisk of my Ultimo 4K. Some people suspect that (again) Lcd4linux might cause this delay so today I removed Lcd4linux. I wasn’t able yet to test extensively.

[Pike_Bishop]

Hi Frenske,

 

at least here Lcd4linux isn't the problem, because i didn't use Lcd4linux.

 

 

regards

Pike

[WanWizard]

To test this, I've created a CIFS mount to my Synology (I always use NFS).  Sofar no issues with the response.

[Pike_Bishop]

Hi WanWizard,

 

thanks, did you use it with autofs ?

and if so did you use nfsvers=4 ?

 

my entry looks as following as i used nfs;

DiskStation	-fstype=nfs,rw,nolock,tcp,nfsvers=4,rsize=8192,wsize=8192,timeo=14,soft	192.168.178.40:/volume1/video

and for samba i use;

DiskStation	-fstype=cifs,vers=3.0,rw,iocharset=utf8,user=xxxxxxx,pass=xxxxxxxxx	://192.168.178.40/video

and as i said , this problem is not always but often, so for test it needs several days, and also a symlink in /media/hdd/movie i think, and we must often open the movielist

in my opinion the problem is coming if the box is longer on.

 

can i debug it ?

 

 

regards

Pike

Edited by Pike_Bishop, 7 October 2019 - 22:54.

[WanWizard]

No, autofs is not standard in OpenPLi. And I know Frenske doesn't use it either.

 

I keep the mount active, we'll see how things progress.

[nor50]

After installing a new Ultimo 4K today, I can verify what I have experied about sub-titles. I wrote earlier here, that on later 6.x or 7., did I get problems with Norwegian subs. Had to activate it on every channel after every update, and the subs was not the same graphical quality as prior. Now with a raw install, it's back to normal.

[mymon]

Just an update about the network share

Found when I have the MS folders mounted via fstab as per below 

//172.16.200.2/C/Users/Admin  /hdd/movie/EeeBox            cifs       rw,username=login,password=password,_netdev            0  0
//172.16.200.2/C/Dropbox      /hdd/movie/DropBox           cifs       rw,username=login,password=password,_netdev            0  0

I can not even lists it via ls command in terminal - just freezes

I've been using this fstab method for years till v7.1 and never had issues (rolled back to 6.2 and no issues)

[WanWizard]

I don't have Windows here, but mount.cifs is a kernel driver, and most boxes haven't changed kernel versions in dogs years...

 

Based on the report above I've created a CIFS mount from my box to my NAS, besides the NFS mount I normally use. And I can't reproduce any of this.

 

I would be interested to see the output of the "mount" command for these two mounts, both from 6.x and 7.x, so we can see if there are any differences in mount option defaults... 

[mymon]

I'll try to share that later on