end_request: critical target error

[Marcel Ribeiro Dantas]

Hi,

Whenever I install dcm4chee, after a few months when it gets over 1TB or so of images, the system stops receiving images.

Dmesg is always flooded with messages like:

end_request: critical target error, dev sdb, sector 6407681024

end_request: critical target error, dev dm-0, sector 6407680000

end_request: critical target error, dev dm-0, sector 6407681024

__ratelimit: 168 callbacks suppressed

sd 4:0:0:1: [sdb] Unhandled sense code

sd 4:0:0:1: [sdb] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE

sd 4:0:0:1: [sdb] Sense Key : Data Protect [current] 

sd 4:0:0:1: [sdb] Add. Sense: Space allocation failed write protect

sd 4:0:0:1: [sdb] CDB: Write(16): 8a 00 00 00 00 01 7d ed 78 00 00 00 04 00 00 00

__ratelimit: 346 callbacks suppressed

sd 4:0:0:1: [sdb] Unhandled sense code

sd 4:0:0:1: [sdb] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE

sd 4:0:0:1: [sdb] Sense Key : Data Protect [current] 

sd 4:0:0:1: [sdb] Add. Sense: Space allocation failed write protect

sd 4:0:0:1: [sdb] CDB: Write(16): 8a 00 00 00 00 01 7d ed 7c 00 00 00 00 20 00 00

__ratelimit: 11738 callbacks suppressed

Buffer I/O error on device dm-0, logical block 800960256

lost page write due to I/O error on dm-0

Buffer I/O error on device dm-0, logical block 800960257

lost page write due to I/O error on dm-0

Buffer I/O error on device dm-0, logical block 800960258

I'm running dcm4chee 2.17.1 with PostgreSQL 9.2.4, JBoss 4.2.3.GA under RedHat Linux 6.3.

I used to have the same configuration in a server and after a few months it happened the same. Then we bought a new server and now it's happening the same thing, which means it's not hardware problem at all.

Running `sudo dumpe2fs /dev/sda2 | more` I get:

Filesystem features:      has_journal ext_attr resize_inode dir_index filetype needs_recovery extent flex_bg sparse_super large_file huge_file uninit_

bg dir_nlink extra_isize

Which means support for over 32.000 subdirectories and everything else is enabled.

What may be happening?

Best regards,

Marcel Ribeiro Dantas,

Biomedical Engineering Researcher

http://mribeirodantas.github.io

ribeirodantasdm at gmail.com

mribeirodantas at lais.huol.ufrn.br

GNU/Linux User nro. #440985