www.ice-graphics.com Forum Index www.ice-graphics.com
The main forum for the ICE-Graphics software
 
 FAQFAQ   SearchSearch   MemberlistMemberlist   UsergroupsUsergroups   RegisterRegister 
 ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

recovery bug in ICEECC 2.0
Goto page 1, 2  Next
 
Post new topic   Reply to topic    www.ice-graphics.com Forum Index -> Bugs and Problems
View previous topic :: View next topic  
Author Message
babaz



Joined: 30 Jul 2005
Posts: 10

PostPosted: Sat Jul 30, 2005 11:53 am    Post subject: recovery bug in ICEECC 2.0 Reply with quote

hi, tnx for this nice apps and tnx for incredibly fast support and updating - now let's go on with the "issue"

i tried a fast one:

a folder with 4 files inside (singular sizes listed below)
1- 13.076.673
2- 24.055.283
3- 2.367.260
4- 12.723.489

i create an ECC set with 10% redundancy and block count=2048
did a FULL TEST --> Full Test POSITIVE -ok

so let's test "real world" recovery...

i completely deleted the file n. 3 above (2.3mb circa)
then i run the recovery
ICEECC says recovery is possible (as i expected) but final process FAILS!
and (interestingly enough) i ended up with this "recovered files":

1- OK
2- CORRUPTED (?)
3- COMPLETELY MISSING (NOT RECREATED AT ALL)
4- CORRUPTED (?)

quite funny uh?
maybe i could try with earlier version (say v1.6) to see if it's a broken feature or a persisting bug...

any clue?
Back to top
View user's profile Send private message
babaz



Joined: 30 Jul 2005
Posts: 10

PostPosted: Sat Jul 30, 2005 12:04 pm    Post subject: Reply with quote

an easy one here...

i reinstalled earlier 1.6 version

i tried to recover files using v2.0 created ECC fileset
(see earlier post above)

guess what?
Recovery Succesfull!

so i assume that latest 2.0 version creates valid ECC sets but has broken recovey routines

trying to help out Wink
Back to top
View user's profile Send private message
ICE Graphics
Site Admin


Joined: 31 Mar 2003
Posts: 430

PostPosted: Sat Jul 30, 2005 8:08 pm    Post subject: Re: recovery bug in ICEECC 2.0 Reply with quote

babaz wrote:
a folder with 4 files inside (singular sizes listed below)
1- 13.076.673
2- 24.055.283
3- 2.367.260
4- 12.723.489

I made the same test. Everything is fine... Try to do it 2-3 times. May be it hardware failure ?


babaz wrote:
quite funny uh?
maybe i could try with earlier version (say v1.6) to see if it's a broken feature or a persisting bug...
any clue?

I think, what it's fail of your computer hardware.

babaz wrote:
i reinstalled earlier 1.6 version

i tried to recover files using v2.0 created ECC fileset
(see earlier post above)
guess what?
Recovery Succesfull!

Recovery parts v2.0 and v1.6 are almost identical.
Back to top
View user's profile Send private message Visit poster's website
babaz



Joined: 30 Jul 2005
Posts: 10

PostPosted: Sun Jul 31, 2005 12:42 am    Post subject: Re: recovery bug in ICEECC 2.0 Reply with quote

ICE Graphics wrote:
babaz wrote:
a folder with 4 files inside (singular sizes listed below)
1- 13.076.673
2- 24.055.283
3- 2.367.260
4- 12.723.489

I made the same test. Everything is fine... Try to do it 2-3 times. May be it hardware failure ?


sorry pal, but it simply cannot be
i've a pretty standard mid-low machine - it's an ECS k7s5a mobo with athlon XP 1700+ and 512mb DDR2100 machine - no overclock of sort
just plain vanilla - and pretty rock solid system (no freeze, no BSODs)
definitely i would not think of hardware failure...

expecially because it's repeteable - would it be an hardware glitch, results would come out mixed or just different and NOT always the same!

so i repeat: i'm pretty confident it's a major flaw in 2.0 version and i ask other users to do some tests (even some basic ones...) before switching completely...

the good part of the story is that even 2.0 created ECC sets seem fine and reliable
that's all Shocked
Back to top
View user's profile Send private message
Glyph



Joined: 05 Jun 2005
Posts: 31

PostPosted: Sun Jul 31, 2005 3:32 am    Post subject: Reply with quote

I could not repeat the failure either. What was the error message? is there something you changed? like the multi-cpu or multi threading features?
Back to top
View user's profile Send private message
ICE Graphics
Site Admin


Joined: 31 Mar 2003
Posts: 430

PostPosted: Sun Jul 31, 2005 7:06 am    Post subject: Re: recovery bug in ICEECC 2.0 Reply with quote

babaz wrote:
sorry pal, but it simply cannot be

Never say never Smile


babaz wrote:
i've a pretty standard mid-low machine - it's an ECS k7s5a mobo with athlon XP 1700+ and 512mb DDR2100 machine - no overclock of sort just plain vanilla - and pretty rock solid system (no freeze, no BSODs) definitely i would not think of hardware failure...

Most program are immune to single RAM or HDD failure. Most program do not not use RAM in stress mode, like ICE ECC do.


babaz wrote:
expecially because it's repeteable - would it be an hardware glitch, results would come out mixed or just different and NOT always the same!

Result can be different. But if recovery result is not equal to original files, ICE ECC will tell always the same: "recovery is fail".


babaz wrote:
so i repeat: i'm pretty confident it's a major flaw in 2.0 version and i ask other users to do some tests (even some basic ones...) before switching completely...

the good part of the story is that even 2.0 created ECC sets seem fine and reliable that's all Shocked

Any bug can be easy fixed, if it's possible to reproduce it. This is bug i can not reproduce. As i know, only hardware failure can not be reproduced...
Back to top
View user's profile Send private message Visit poster's website
babaz



Joined: 30 Jul 2005
Posts: 10

PostPosted: Sun Jul 31, 2005 10:53 am    Post subject: Reply with quote

Glyph wrote:
I could not repeat the failure either. What was the error message? is there something you changed? like the multi-cpu or multi threading features?


everything was left at default values - so multithreading ON and MultiCPU 1 (of course, since it's just an athlonXP 1700+)

the error repoted "defective block **" and then printed out the name of one of the .ECC files (i forgot to say that i created a "power of 2" set of ECC files.. not just 1 single ECC file!)

i'm going to redo exactly that test today - let's see if my RAM fails always in the same way Very Happy

no pun intended here, but i think there are a lot of other programs that stress RAM quite a bit (say file compressors like WinRAR or 7zip - which i use every day - or media compressors like LAME or FLAC < since this is lossless i could encode and decode MANY times and always come up with the same .wav file - strange uh?)
Back to top
View user's profile Send private message
babaz



Joined: 30 Jul 2005
Posts: 10

PostPosted: Sun Jul 31, 2005 12:06 pm    Post subject: Reply with quote

OK - i did the same test again with v2.0
no go, as i expected

here's my log for ECC power of 2 sizing scheme creation (ok), FULL ECC check (ok) and subsequent recovery (fail)
i've trimmed / censored something for privacy Very Happy

Quote:

ICE ECC v2.0
Physical Memory - 511 MB

Speed - 373,73 MB/Sec
Configuration - 13, 0, 4

Create started: D:\roba da alex!; D:\roba da alex!\roba da alex!.ecc
Matrix calculation
Creating ECC Files
Matrix calculation
Finalization
Time - 0:00:44.34
Process finished
Create finished

Full ECC Test started: D:\roba da alex!\;
Check - roba da alex! 1-1.ecc
--skip--
Test header
Header Redundancy - 1600%
Test ECC Files
All files are valid
Full ECC Test finished

Verify only started: D:\roba da alex!; D:\roba da alex!\roba da alex! 1-1.ecc
Check - roba da alex! 1-1.ecc
--skip--
Verify Data Files
File read error - D:\roba da alex!\--skip--.7z
Damaged block - 1457
--skip--
File read error - D:\roba da alex!\--skip--.7z
Damaged block - 1549
Verify ECC Files
Some files are damaged. Recover is possible
Verify only failed
Select ECC file(s)

Verify started: D:;
Check - roba da alex! 1-1.ecc
--skip--
Verify Data Files
File read error - D:\roba da alex!\--skip--.7z
Damaged block - 1457
--skip--
File read error - D:\roba da alex!\--skip--.7z
Damaged block - 1549
Verify ECC Files
Some files are damaged. Recover is possible
Matrix calculation - 0
Recovery Data Files
Damaged block - 30 - D:\roba da alex!\roba da alex! 5-16.ecc
Recover failed

Verify failed

then again, if my RAM is at fault why IceECC 1.6 works (apparently) flawlesslly?
Back to top
View user's profile Send private message
ICE Graphics
Site Admin


Joined: 31 Mar 2003
Posts: 430

PostPosted: Sun Jul 31, 2005 1:35 pm    Post subject: Reply with quote

babaz wrote:
OK - i did the same test again with v2.0
no go, as i expected

here's my log for ECC power of 2 sizing scheme creation (ok), FULL ECC check (ok) and subsequent recovery (fail)
i've trimmed / censored something for privacy Very Happy


You can see my log:
Code:
ICE ECC v2.0
Physical Memory - 2047 MB

Speed - 744,06 MB/Sec
Configuration - 4, 8, 6

Verify started: F:\A;
Check - A 4-8.ecc
Use ECC file - F:\B\A 4-8.ecc
Check - A 1-1.ecc
Use ECC file - F:\B\A 1-1.ecc
Check - A 2-2.ecc
Use ECC file - F:\B\A 2-2.ecc
Check - A 3-4.ecc
Use ECC file - F:\B\A 3-4.ecc
Check - A 5-16.ecc
Use ECC file - F:\B\A 5-16.ecc
Check - A 6-32.ecc
Use ECC file - F:\B\A 6-32.ecc
Check - A 7-64.ecc
Use ECC file - F:\B\A 7-64.ecc
Check - A 8-77.ecc
Use ECC file - F:\B\A 8-77.ecc
Verify Data Files
File read error - F:\A\0002
Damaged block - 1457
File read error - F:\A\0002
Damaged block - 1458
-- Skip --
File read error - F:\A\0002
Damaged block - 1548
File read error - F:\A\0002
Damaged block - 1549
Verify ECC Files
Some files are damaged. Recover is possible
Matrix calculation - 0
Recovery Data Files
Finalization
All files succesfully recovered
Verify finished



babaz wrote:
then again, if my RAM is at fault why IceECC 1.6 works (apparently) flawlesslly?

However v1.6 and v2.0 use the same sources, binary files (EXE) always are different. Besides, there are some differences in code. Different versions occupy different addresses in RAM.

Did you try to make other tests? Other test data? Recover always fail, even if only one block is damaged? Did you try to test on other computer?
Back to top
View user's profile Send private message Visit poster's website
Glyph



Joined: 05 Jun 2005
Posts: 31

PostPosted: Sun Jul 31, 2005 1:50 pm    Post subject: Reply with quote

To babaz:

I know this might sound cumbersome: could ya find a small data set (say only 1 meg in size) that also produces the same error and then send it to me for testing? I got several different computers here, and would like to try and reproduce the error.

As for privacy, i'm sure you can find 1 meg worth of say redundant windows files or movie trailer bits to test it on.

Be sure to include the parameters of the test.

you can send it to:

Rayjing@hotmail.com
Back to top
View user's profile Send private message
babaz



Joined: 30 Jul 2005
Posts: 10

PostPosted: Sun Jul 31, 2005 3:23 pm    Post subject: Reply with quote

ICE Graphics wrote:
babaz wrote:
OK - i did the same test again with v2.0
no go, as i expected

here's my log for ECC power of 2 sizing scheme creation (ok), FULL ECC check (ok) and subsequent recovery (fail)
i've trimmed / censored something for privacy Very Happy


You can see my log:
--cut--


of course i understand that ur system is working...
but your system is pretty different to mine, just look at those 2GB RAM (which is surely no mainstream setup, uh?)

ICE Graphics wrote:

babaz wrote:
then again, if my RAM is at fault why IceECC 1.6 works (apparently) flawlesslly?

However v1.6 and v2.0 use the same sources, binary files (EXE) always are different. Besides, there are some differences in code. Different versions occupy different addresses in RAM.


ok, that makes sense: but still would it be that i'm so unlucky that v2.0 goes triggering just those faulty RAM areas [taking for grant that i have any faluty RAM at all... which i don't believe] ?!?! c'mon, that's not realistically possible

ICE Graphics wrote:

Did you try to make other tests? Other test data? Recover always fail, even if only one block is damaged? Did you try to test on other computer?


yes done other tests > failed!
yes i've used another set of data > failed!
recover FAILS when i delete the file and FAILS even if i corrupt just 1 block
no, ATM i've not tried on different computers

but still, please consider this: v1.6 works as supposed
so, where's the bug? in my RAM? is everybody really taking this seriously?

please forgive my "polemic" approach, english is not my native language

EDIT: the v2.0 tests were done using .zip distro, whereas v1.6 tests were done using .exe installer (just for the sake of it)


Last edited by babaz on Sun Jul 31, 2005 3:28 pm; edited 1 time in total
Back to top
View user's profile Send private message
babaz



Joined: 30 Jul 2005
Posts: 10

PostPosted: Sun Jul 31, 2005 3:26 pm    Post subject: Reply with quote

Glyph wrote:
To babaz:

I know this might sound cumbersome: could ya find a small data set (say only 1 meg in size) that also produces the same error and then send it to me for testing? I got several different computers here, and would like to try and reproduce the error.

As for privacy, i'm sure you can find 1 meg worth of say redundant windows files or movie trailer bits to test it on.

Be sure to include the parameters of the test.

you can send it to:

Rayjing@hotmail.com


tnx, i'll try to setup one new test-bed right away (boring sunday u know...)
Back to top
View user's profile Send private message
ICE Graphics
Site Admin


Joined: 31 Mar 2003
Posts: 430

PostPosted: Sun Jul 31, 2005 6:24 pm    Post subject: Reply with quote

babaz wrote:
of course i understand that ur system is working...
but your system is pretty different to mine, just look at those 2GB RAM (which is surely no mainstream setup, uh?)

Memory size is not important in your case because size of test data only 50MB.

babaz wrote:
but still, please consider this: v1.6 works as supposed
so, where's the bug? in my RAM? is everybody really taking this seriously?

I hope test with your data on other computer can help you to understand who force ICE ECC recover to fail. Is it software bug or hardware failure.

P.S. I know what it's situation annoy you. But you have to understand me. I can help only if i can reproduce problem.
Back to top
View user's profile Send private message Visit poster's website
ICE Graphics
Site Admin


Joined: 31 Mar 2003
Posts: 430

PostPosted: Sun Jul 31, 2005 7:09 pm    Post subject: Reply with quote

Problem fixed. It's a bug. Bug was linked with recovering in single CPU mode. v2.0a is online.
Back to top
View user's profile Send private message Visit poster's website
Glyph



Joined: 05 Jun 2005
Posts: 31

PostPosted: Sun Jul 31, 2005 7:33 pm    Post subject: Tested and working Reply with quote

Looks good.

I was able to reproduce the error using the dataset Babaz gave me with ICE ECC 2.0

It is indeed something to do with the multi-cpu feature. I didn't notice it before because i was always using 2 CPUs since i have a dual system.

The problem does seem to be fixed in 2.0a for both 1 and 2 CPU configurations.


So everything looks fine now, at least on my end. Very Happy

I don't have 3 or 4 cpu system so i can't test it fully in that way.
Back to top
View user's profile Send private message
Display posts from previous:   
Post new topic   Reply to topic    www.ice-graphics.com Forum Index -> Bugs and Problems All times are GMT
Goto page 1, 2  Next
Page 1 of 2

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Powered by phpBB © 2001, 2005 phpBB Group