Jump to content
Corsair Community

Problem with Voyager 64G


jari

Recommended Posts

I'm having some problems with Voyager 64G. I have not been able to use it since I do not trust it will function as should i.e. the behavior will cause corruption in file system.

 

Test:

 

Truecrypt 6.0a creating 35Gb encrypted file or partition to ext2/ntfs partition on sdb1 device in linux.

 

(Formating drive with any of ext2/ntfs/fat32 works ok but creating large file does not)

 

Behavior:

 

* First slowing down in write performance: 6.8Gb

* next 8.4Gb

* next 13.0Gb (long one)

* average transfer speed lowered to 7.6Mb/s

* next 13.2Gb,

* average transfer speed fluctuating between 7.4-7.5Mb/s

* next 15Gb, transfer down to 7.3Mb/s

* next 16Gb, transfer down to 7.2Mb/s

* next 17Gb, transfer down to 7.1Mb/s

* next 18Gb, transfer down to 7.0Mb/s

* next 20Gb, transfer down to 6.9Mb/s

* next 23Gb, transfer down to 6.5Mb/s

* very long transfer spped fluctuation around 1Mb/s

* again with 24Gb

* very long fluctuation around 1Mb/s, transfer speed very long around 1Mb/s (visually examined) and average started to decrade rapidly (of course)

* at 26Gb transfer started to rise again...

* at 27Gb transfer fluctuation around 1Mb

* at 28Gb transfer started to rise again...

* at 31Gb around 1Mb/s again

 

Test system: Mandriva Linux 2009.0 (64bit) with self compiled kernels from http://www.kernel.org source 2.6.27.7, 2.6.27.10, 2.6.28, Core2 optimization turned on. Lenove Thinkpad X61t Dual Core2 with 4Gb memory.

 

During stress test following can be found in /var/log/messages:

 

attempt to access beyond end of device
sdb1: rw=1, want=24241084240, limit=126353376
attempt to access beyond end of device
sdb1: rw=1, want=28673439160, limit=126353376
attempt to access beyond end of device
sdb1: rw=1, want=19584375208, limit=126353376
attempt to access beyond end of device
sdb1: rw=1, want=28066738672, limit=126353376
attempt to access beyond end of device
sdb1: rw=1, want=28042160152, limit=126353376

 

and:

 

Dec  1 21:51:08 dumbo kernel: __ratelimit: 32020 callbacks suppressed
Dec  1 21:51:08 dumbo kernel: iwlagn: Can not allocate SKB buffers
Dec  1 21:51:15 dumbo last message repeated 9 times
Dec  1 21:51:24 dumbo kernel: __ratelimit: 1254953 callbacks suppressed
Dec  1 21:51:24 dumbo kernel: iwlagn: Can not allocate SKB buffers
Dec  1 21:51:48 dumbo last message repeated 12 times

usb 2-1: USB disconnect, address 10
sd 17:0:0:0: [sdb] Result: hostbyte=DID_NO_CONNECT driverbyte=DRIVER_OK,SUGGEST_OK
end_request: I/O error, dev sdb, sector 25300608
Buffer I/O error on device sdb1, logical block 3162572
lost page write due to I/O error on sdb1
...<lines removed>...
Buffer I/O error on device sdb1, logical block 3173554
lost page write due to I/O error on sdb1
sd 17:0:0:0: [sdb] Result: hostbyte=DID_NO_CONNECT driverbyte=DRIVER_OK,SUGGEST_OK
end_request: I/O error, dev sdb, sector 25330992
sd 17:0:0:0: [sdb] READ CAPACITY failed
sd 17:0:0:0: [sdb] Result: hostbyte=DID_NO_CONNECT driverbyte=DRIVER_OK,SUGGEST_OK
sd 17:0:0:0: [sdb] Sense not available.
sd 17:0:0:0: [sdb] Write Protect is off
sd 17:0:0:0: [sdb] Mode Sense: 00 00 00 00
sd 17:0:0:0: [sdb] Assuming drive cache: write through

 

Same problems happen if I try to create 35Gb partition (/dev/sdb2) to USB stick.

 

My interpretation of the problem is that for some reason usb stick will disconnect when certain stress is applied to it. Machine port does not affect the behavior, neither does computer (tested with 2, the other was Dual Core AMD).

 

I can format the disk under WindowsXP and create the Truecrypt file but the same write behavior is still visible. I have tried to mimic Windows by setting the "echo 15 >/sys/module/scsi_mod/parameters/inq_timeout" instructed in http://www.houseofhelp.com/forums/showthread.php?t=73036. This does not help or affect the outcome of the test. I do not think this is Linux kernel problem rather that it is just better visible under Linux. The problem is most likely in the Flash drive.

 

Any ideas to correct the problem appreciated... Stick was quite expensive.

Link to comment
Share on other sites

  • Corsair Employees

First thing

You need to understand some limitations of the Fat 32 File system.

And when using the drive in a non Windows system it may not see or function properly

Fat32 File Limitations Wiki Pedia

FAT32 File System in Windows XP

Limitations of FAT32 File System

 

In a nut shell you will not be able to use the full size with fat 32 above 32 Gig when using other O.S. other than Windows.

If you will only use Windows XP there should not be any problems, but if you will use Linux you may have to only use Linux and I would not suggest using encryption as the limitation will be strictly applied the way I understand the Implementation of the Fat file system in any O.S. but especially with Linux and Mac O.S..

Link to comment
Share on other sites

What do you mean by "limitations of FAT32" filesystem? The drive is not formated as FAT32 anymore. As I said it does not matter if the drive is formated as ext2/ntfs or fat32. Behavior is the same. Of course when it is formated as fat32 the drive cannot hold 32Gb file but with ext2 or ntfs this is be possible. However, creating 16 2Gb files to fat32 formated Voyager causes save problem.

 

Different test cases were (voyager==sdb, partition 1 on voyager==sdb1, partition 2 on voyager==sdb2, filesystem on truecrypt volume can be whatever).

 

Test 1:

* sdb1: ntfs, 64Gb. Test was trying to create 35Gb Truecrypt volume as single file.

Test 2:

* sdb1: ext2, 64Gb. Test was trying to create 35Gb Truecrypt volume as single file.

Test 3:

* sdb1: ntfs, 29Gb.

* sdb2: truecrypt volume, ~35Gb.

Test 4:

* sdb1: ext2, 29Gb

* sdb2: truecrypt volume, ~35Gb.

Test 5:

* sdb1: fat32, 29Gb

* sdb2: truecrypt volume, ~35Gb.

Test 6:

* sdb1: fat32, 64Gb. Creating 16 2Gb files onto the drive.

 

All tests cause the behavior described in my 1st post.

 

Formating the drive into any of ext2/fat32/ntfs was not a problem (unless I used bad block chech, which causes same problematic behavior). Creating the truecrypt volume as separate partition or file volume failed - as did any bad block verification of the flash drive.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...