69 replies to this topic

[mman]

Do I need to flash or just do software updates from my box, is the update OK now????

[dirocca]

you need to flash with image of today

 

grtz Philip

[Lost in Space]

When you just do an upgrade you would have latest kernel and drivers BUT still not ubifs. BUT as somebody 'forgot' that UBIFS has lzo compression as default but jffs2 has zlib and PLI used lzo for jffs2 ... then your old jffs2 filesystem will be not mountable anymore by the new kernel and the box will stop booting.

 

Dreambox 8000/7020hd owners with standard OE 2.0 based images could use Marsu Plugin for UBIFS conversion, or now dFlash 9.4 - and the later should work also with PLi Images as the Marsupilami is now back to the jungle.

 

If you don't like this backup/restore approach the only choice to get ubifs is flashing an up-to-date image.

 

For the other boxes (500hd/800se) the backup/restore conversion is still the only way to get ubifs.

Edited by gutemine, 20 April 2013 - 17:32.

[mman]

So as the owner of DM 8000 from now on I can't do software update everytime I will need to flash?

[Lost in Space]

NO, as soon you have done the Filesystemswitch you can update as normal.

[mman]

Never done that. Sorry for not knowing but how do I do Filesystemswicth?

[Meega]

Never done that. Sorry for not knowing but how do I do Filesystemswicth?

Just by flashing the newest build

[dirocca]

Never done that. Sorry for not knowing but how do I do Filesystemswicth?

he means flash

 

grtz Philip

[mman]

OK Thanx everyone

[Lost in Space]

As I wrote already - If you don't want to flash a vergin image but keep all your existing image content OE 2.0 users could upgrade their box smoothly and then do a backup with ubifs and restore it ot get ubifs.

 

The same procedure for the PLi images was NOT tested at all (congratulations!) and seems to be not yet fixed either.

 

opkg install --force-reinstall dreambox-secondstage-dm7020hd
Removing package dreambox-secondstage-dm7020hd from root...
Installing dreambox-secondstage-dm7020hd (87-r14.0) to root...
Downloading http://downloads.pli...0_dm7020hd.ipk.
Configuring dreambox-secondstage-dm7020hd.
/usr/share/dreambox-secondstage-dm7020hd/secondstage-dm7020hd.bin: No such file or directory
Invalid boot partition!
Erased 1048576 bytes from address 0x00000000 in flash
Writing data to block 0 at offset 0x0
 

Well done, now the erase part of the SSL install script erased the old ssl, but there was no new ssl file found as the path is wrong.

 

/usr/share/dreambox-secondstage-dm7020hd/secondstage-dm7020hd.bin: No such file or directory

root@dm7020hd:/media/hdd# ls -al /usr/share/dreambox-secondstage
drwxr-xr-x    2 root     root           248 Jan  1 08:34 .
drwxr-xr-x   24 root     root          2000 Jan  1 08:34 ..
-rw-r--r--    1 root     root        144464 Apr 19  2013 secondstage-dm7020hd.bin

 

So instead of fixing such simple things on the feed you caused yourself, you are telling the people better to flash their box (with DreamUP as now the SSL is dead)
 

Edited by gutemine, 20 April 2013 - 19:11.

[Meega]

As I wrote already - If you don't want to flash a vergin image but keep all your existing image content OE 2.0 users could upgrade their box smoothly and then do a backup with ubifs and restore it ot get ubifs.

 

The the same procedure for the PLi images was NOT tested at all (congratulations!) and seems to be not yet fixed either.

 

opkg install --force-reinstall dreambox-secondstage-dm7020hd
Removing package dreambox-secondstage-dm7020hd from root...
Installing dreambox-secondstage-dm7020hd (87-r14.0) to root...
Downloading http://downloads.pli...0_dm7020hd.ipk.
Configuring dreambox-secondstage-dm7020hd.
/usr/share/dreambox-secondstage-dm7020hd/secondstage-dm7020hd.bin: No such file or directory
Invalid boot partition!
Erased 1048576 bytes from address 0x00000000 in flash
Writing data to block 0 at offset 0x0
 

Well done, now the erase part of the SSL install script erased the old ssl, but there was no new ssl file found as the path is wrong.

 

/usr/share/dreambox-secondstage-dm7020hd/secondstage-dm7020hd.bin: No such file or directory

root@dm7020hd:/media/hdd# ls -al /usr/share/dreambox-secondstage
drwxr-xr-x    2 root     root           248 Jan  1 08:34 .
drwxr-xr-x   24 root     root          2000 Jan  1 08:34 ..
-rw-r--r--    1 root     root        144464 Apr 19  2013 secondstage-dm7020hd.bin

 

So instead of fixing such simple things on the feed you caused yourself, you are telling the people better to flash their box.
 

Thought this topic is about a dm8000?

I have one and flashed the box without problems.

[Lost in Space]

Yes it is, but the trouble is the same and I'm not willing to install PLi on all my boxes when you do such great quality verification :-)

 

And it is not about flashing a new image, the above example shows what happens on an upgrade by forcing a loader re-install which eareses the loader but fails to write the new one.

Edited by gutemine, 20 April 2013 - 19:17.

[Meega]

Yes it is, but the trouble is the same and I'm not willing to install PLi on all my boxes when you do such great quality verification :-)

 

And it is not about flashing a new image, the above example shows what happens on an upgrade by forcing a loader re-install which eareses the loader but fails to write the new one.

ah,but mman could flash (if he has a dm8000) and he will be fine.

[Lost in Space]

Yes off course, because then the SSL will come from the nfi image and the opkg status database the already contains the new package version and hence this problems occures only when you force a reinstall of the ipk as my output shows.

 

BUT if you do an upgrade from a previous image this happens during the upgrade and then you go for the nullmodem cable because of such a simple to fix error.

 

Look I just wanna try to help fixing it, it was just a little strange that such an obvious error is not fixed within 5min and instead everybody telsl the people to flash the new image. The recomandation is OK on the 8000 and 7020hd as he the will also get the new UBIFS filesystem BUT why let him first grill his box when he does a Softwareupdate? And as i said this doesn't fix the error, it just avoids runing into it.

 

If it would have been properly implemented without this error in the postinst script of the loader packages then the backup/restore upgrade path for the filesystem would work also as I explained already.

 

Especially on the 500hd and 800se where there is NO ubifs the argument to flash just to overcome this error is even more annoying.

Edited by gutemine, 20 April 2013 - 19:43.

[Meega]

Oke, i understand what you mean.

[Lost in Space]

Well, I was recently sold because I don't do all testing by myself, so maybe my approach is simply not that bad as I was told

Edited by gutemine, 20 April 2013 - 19:51.

[pieterg]

For those of you who are new to our system of nightly OpenPLi builds;
Of course there are times when we broke something, or have work in progress.
When that happens, we put the online updates on hold.
Users get a big warning when they try to update, if the nightly build is not considered safe at that time.
If they choose, and know what they are doing, they can force the update to continue if they want.
But generally, people take the warning seriously, and wait for the 'all clear'.

So if you notice the download images for certain boxes are red, with a big skull icon, and you receive a warning during the online update, you know what's going on.

In this particular case, as has been explained on the forum in several threads over the last few days, it is not safe to do an online update, but it is safe to flash the latest image.

[Lost in Space]

I fully understand this approach of try and errror (as I use a similar one which has proven to be prette efficient) but when images are safe and updates a problem putting a skull to the image is .... confusing.

 

Even when it is axplained in lots of threads.

 

And things can always go wrong, especially on such complex changes, but as I pointed out the error here is easy to find and easy to fix so the only test is our patience :-)

Edited by gutemine, 20 April 2013 - 21:35.

[Lost in Space]

Thanks for fixing the loader update problem.

Edited by gutemine, 21 April 2013 - 09:06.

[littlesat]

When there is a skul on the website... The upgrade process via the remote gives you a big warning and referred to our website... What is confusing about this?