Jump to content

Problems with apt-get upgrade on cubietruck


chrissy_D

Recommended Posts

Hi,

 

I have serious problems with the 5.25 images of armbian on a cubietruck. After a fresh install and apt-get update apt-get upgrade I get errors with :

 

dpkg: error processing archive /var/cache/apt/archives/linux-libc-dev_4.4.0-78.99_armhf.deb (--unpack)

 

This is for the Armbian_5.25_Cubietruck_Ubuntu_xenial_next_4.9.7 image, with other images this concerns a different package. Only Armbian_5.25_Cubietruck_Ubuntu_xenial_default_3.4.113 works.

After this error, the filesystem seems to be defect and is remounted ro. On another try, I removed the rmount ro on error option from the fstab and the /var/lib/dpkg directory was corrupted.

I tried several new SD cards and two cubietrucks, result is still the same.

 

Any help appreciated.

Greets

Christian

 

Link to comment
Share on other sites

7 minutes ago, chrissy_D said:

I tried several new SD cards and two cubietrucks, result is still the same.

 

Since you didn't mention that you replaced the power supply also this would be next check after testing at least one of your new SD cards with either F3 or H2testw (put the logs on pastebin.com or show H2testw screenshot here for review please) and providing the CRC32 checksum Etcher shows you after successfully burning the OS image. This looks like this for example:

Bildschirmfoto%202017-05-22%20um%2016.22

Link to comment
Share on other sites

Hi,

I tried different powersupplies too, same result. Today I checked one SD-Card with  H2TESTW :

 

Achtung: Nur 30524 von 30525 MByte getestet.
Fertig, kein Fehler aufgetreten.
Sie können die Testdateien *.h2w jetzt löschen oder nach Belieben
nochmals überprüfen.
Schreibrate: 16,1 MByte/s
Leserate: 26,6 MByte/s
H2testw v1.4

 

eg everything is okay.

 

Flashed the image with etcher instead of Win32DiskImager, flash completed flawlessly. Booted cubietruck with the same image. Same error after apt-get update, apt-upgrade at libc-dev.

 

No Ideas anymore.

 

Greets

Christian

 

PS: I downloaded the latest etcher version today (v1.0.0 - 2017-05-12)

 

etcher_flash.png

Link to comment
Share on other sites

Thank you for ruling out the two 'standard' problems. Can you please post output from 'sudo armbianmonitor -u' if your still running the image (otherwise please reflash the image again, login, finish user creation and post then the output before and after 'apt upgrade')

Link to comment
Share on other sites

2 hours ago, chrissy_D said:

Hmm... really no idea what's going on but you end up with a corrupted filesystem regardless what you do as soon as write to the FS on the Cubietruck. Just a quick note: if you search for 'quick iozone test' in the sprunge link and compare with EVO numbers here then it looks your card is pretty slow (mmc metadata seems to be genuine).

Link to comment
Share on other sites

1 hour ago, chrissy_D said:

Both cards I tried are new Samsung EVO+ 32 GB Cards.

 

I know since that's what '### mmc' section in http://sprunge.us/fLFL already told me. I just wondered why those cards are that slow. Are you able to run flashbench on one of the cards when in a card reader connected to a PC and provide results? Or even a CrystalDiskMark screenshot (me interested in 4K/QD32 numbers)?

Link to comment
Share on other sites

Hi,

 

its very strange, today I tried an ol 8GB SD Card by sandisk and everything worked fine. I guess both new Samsung cards are corrupted in a manner, that even the test programs don't recognize it. Strange but seems to be true. Too bad I dumped the invoice, so I can't return them.

 

But it is definitively no flaw of the image.

 

Greets Chris

Link to comment
Share on other sites

6 minutes ago, chrissy_D said:

But it is definitively no flaw of the image.

Of course, these images are downloaded all the time and if they are corrupted or introduce problems users report this very quickly and this would've been fixed long ago.

 

The reason I was asking for CrystalDiskMark 4K/QD32 numbers for one of these cards is that the write pattern is similar to what happens when you run an 'apt upgrade' (a lot of random writes). If you're able to trigger same symptoms you should ask for a refund anyway.

 

I don't recall where I read this recently but someone reported a 'new era' of counterfeit SD cards. They do not fake capacity any more (report 128 GB but in reality just 16 GB for example) but are just utter crap surviving the F3 or H2testw test runs (sequential write/read) but die then later when used more heavily.

Link to comment
Share on other sites

Hi,

 

here is the CrystalDiskMark resultof both Cards. I Bought them in a large electronics strore in Düsseldorf (Conrad Electronics, if you know it) so I don't think they sell counterfeited shit, but, shit happens.

CrystalDiskMark_Samsung_EVO.png

CrystalDiskMark_Samsung_EVO2.png

Link to comment
Share on other sites

2 hours ago, chrissy_D said:

I Bought them in a large electronics strore in Düsseldorf (Conrad Electronics, if you know it) so I don't think they sell counterfeited shit, but, shit happens.

It's not about 'evil sellers' but counterfeit cards being inserted into the supply chain pretty early. You get them everywhere (just do a google search for 'counterfeit sd card samsung site:amazon.com' or similar) and your only hope is (was) checking them immediately which is of course not sufficient any more if now counterfeit cards appear that do not fake capacity but look like genuine cards, show the capacity of genuine cards but are just fakes of lower quality.

 

Thanks for the numbers (still doesn't look like genuine Samsung based on performance but how to tell for sure? Maybe only buy at retailers that refund without asking questions?)

 

Further readings:

It seems in the meantime some of the ways to deal with this problem don't work any more :( 

Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
×
×
  • Create New...

Important Information

Terms of Use - Privacy Policy - Guidelines