[1.0RC3] XBMC keeps restarting

[mk01]

(16th Dec, 2014 05:43 PM)mk01 Wrote:  anyhow with menakite we implemented a fix today. xbmc needs to be recompiled - any xbmc(rpi/imx6) or kodi(rpi/imx6) with suffix higher than -1418714780 (today) should have the fix included (without any additional job conf modding etc).

if anyone install new package and will still feel this issue be present, let us know.

I should have had been more precise - on that - the XBMC part was LAST of three fixes in three different packages

xbian-package-xbmc
xbian-package-upstart-bridge
xbian-package-xbmc-scripts

missing ANY of the 'chain' can retrigger/not solve the problem. but the other two packages been fixed & pushed before xbmc - developer (in that case myself) expects somehow automatically that other will have it - what of course mostly is not true.

so - so - lets correct / reconsolidate the info:

xbian-package-xbmc-scripts at minimum 1.1.4
xbian-package-upstart-xbmc-bridge 1.1.6
*xbmc* (that includes plain xbmc package, suffixed -6q, suffixed -nighly, suffixed -nightly-6q) with numbered suffix HIGHER than the date (16th Dec, 2014 08:43 AM) what translates to:

Code:

date --date="Dec 16, 2014 08:43" +%s

so the number suffix MUST be higher than this 1418715780.

this is full info.
also unfortunately all that mixes with the repo problem we had & restructuring it as part of recovery process.
so to put that as answer to @palswim

new structure is the way like Debian defined it - meaning platforms are horizontal dimension - meaning specifically:

Code:

deb mirror://apt.xbian.org/mirror.txt      stable  main imx6-jessie
deb mirror://apt.xbian.org/mirror.txt     staging   main    imx6-jessie
deb mirror://apt.xbian.org/mirror.txt     devel   main    imx6-jessie

(THIS IS EXAMPLE FROM IMX6, RPI USERS MUST HAVE RPI-WHEEZY instead of IMX6-JESSIE)

also the specs "stable main imx6-jessie" doesn't mean 1st OR 2nd OR 3rd. correct ENTRY for STABLE repo must contain ALL three levels/specs in the line - because only then it makes meaning -

Code:

packages considered STABLE
installable to all platforms so from MAIN (shared packages)
and also the platform specific for IMX6-JESSIE

missing any of the levels will WORK technically, but will never distribute full scale of what needs to be distributed.
to HAVE the correct entries there - user doesn't need to retype all that info manually ! XBIAN-PACKAGE-REPO since version 1.0.6-1 WILL DO THAT CORRECTLY (ALSO AUTOMATICALLY CONSIDERING platform and BASE OS) - but all the info is in the FAQ of http://www.xbian.org

ALSO BOTH files /etc/apt/sources.list and /etc/apt/sources.list.d/xbian.list MUST exists. the first one we are leaving completely for system (and user) - so that defines the upstream debian/raspbian sources - 99% of packages in system comes from there. the rest is coming from xbian repos defined in xbian.list.
ALSO three lines (stable,staging,devel) is provided AS template. currently for RPI only STABLE is requirement. for IMX6 STABLE+STAGING is requirement due to the nature of DEBIAN/Jessie itself - not still stable release.
"Requirement MEANS the line MUST NOT BE commented - starting with #"
@palswim
is this clear now? does it make sense ?

now - all that text - its purpose is not any blame to anyone - it is trying to say that natural way of communication works EXACTLY as defined by other philosophers centuries ago. from my brain (100% of info) - only 80% gets passed to 1st user, 60% to 2nd, 40% to third. that's why - if there is at least ONE user not understanding the message, the message wasn't said/written correctly (yes, my case mostly ;-)). but the user needs to ASK, to tell US. there is no shame in that. only if we are told (like someone here told - he doesn't get it from the FAQ) - only then we can change it. TELL US - there are not stupid questions - only stupid answers.
I will rather (and happier) be answering 3x in a row "YES, YOU UNDERSTAND IT CORRECTLY", comparing to for instance situation with trying to fix a BUGXYZ to find at the end that I created it myself BY improper FAQ entry or Forum message not being detailed enough. so AGAIN, if you-user don't get the text, ASK, TELL us. and be sure, you immediately helped 100 others who don't ask or simply skip that info as it makes no sense.

It will happen regularly that we - people around project - will loose the right user vocabulary - because we understand each other so we forget now and then that it doesn't work for user. TELL me, or us - that hey man! this info was not much useful from user perspective - ask for rewrite or call it rubbish. tell the author that he is slowly sliding off into the tech zone. author might not be AWARE of that.

back to xbmc looping:
- if the above conditions (the three RIGHT versions of debs ARE installed on your XBian and you still have XBMC looping - try to change TIMEOUT=100 to TIMEOUT=99999 on line 43 of /etc/init/xbmc.conf . few days BACK i have seen PROPER UNBROKEN RPI machine running XBIan & all the nice downloader packages etc which was starting XBMC for 3.5 minutes! it was also mounting 3TB disks, file indexes etc.
in the last version of xbian-package-xbmc-scripts we changed from 100s to 150s. what is 2minutes and 30s. 99% of users MUST be below that (maybe we even may move the VALUE into /etc/default/xbmc to be user editable and script will load it from there.

But let's consider the fact that this timeout=100 is there since XBian Beta2. more than 1Y of usage behind. the actual loops have been caused by combination of two other bugs introduced - together causing this FAIL to start). I'm confident about the code again - still if the three debs updates DO NOT SOLVE the issue - report to this thread and I will personally go user by user to track what ever else could by the issue.
Also because never before this happened at MASS scale - we neither users ever considered the looping itself for a BUG. the looping is alone a FEATURE - to auto restart XBMC in case of crash. but if XBMC doesn't start AT ALL (to HOME SCREEN) as part of BOOT and is killed on timeout - it is not logical to try loop it again and again. only today I named as BUG and fixed in xbian-package-xbmc-scripts v1.1.5.
(i have the feeling @palswim challenged that looping few msgs back as well -and clearly - reload AFTER xbmc running & crashing later is the feature we want to havel. useless restarting of XBMC which never make it to display on TV is an attack on user.

last but not least !
I WISH ALL the BEST in 2015 to all fans/users/coworkers/anyone involved !