## page was renamed from ZEOM/Paris/Bureautique/Maintenance/20080808

'''Incident serveur Babel''' (Transtec)  http://wiki.auf.org/wikiteki/ZEOM/Paris/Bureautique/Babel

'''Date''' : vendredi 08/08/2008 

'''Statut''': Fermé

'''Description du problème :''' Plantage complet du serveur

{{{
BUG: unable to handle kernel paging request at virtual address 67e7c4b9
  printing eip:
f89089bd
*pde = 00000000
Oops: 0002 [#1]
Modules linked in: button ac battery ipv6 dm_snapshot dm_mirror dm_mod 
loop rtc psmouse serio_raw floppy shpchp pci_hotplug pcspkr i2c_nforce2 
i2c_core evdev ext3 jbd sd_mod ide_cd cdrom amd74xx generic ide_core 
3w_9xxx forcedeth sata_nv ehci_hcd ohci_hcd libata scsi_mod usbcore 
thermal processor fan
CPU:    0
EIP:    0060:[<f89089bd>]    Not tainted VLI
EFLAGS: 00010286   (2.6.18-6-486 #1)
EIP is at journal_commit_transaction+0x22d/0xc08 [jbd]
eax: 00000000   ebx: e1c1fc40   ecx: c62a4b64   edx: 000000d4
esi: 67e7c489   edi: c40da1d8   ebp: dfad8380   esp: f7819f3c
ds: 007b   es: 007b   ss: 0068
Process kjournald (pid: 1817, ti=f7818000 task=dff92570 task.ti=f7818000)
Stack: 00000000 f7ceb800 00000000 00000000 c7665174 dfdc2740 dff92570 
c0360454
        f7ceb800 000000d4 00000001 00000046 00000000 0000000a dffa0570 
d757ea8e
        000019ad 000067ee dff92680 dfad83d0 f7819fa0 1fd084eb d9eaff74 
dfad8380
Call Trace:
  [<f890bca1>] kjournald+0x92/0x184 [jbd]
  [<c0122cfb>] autoremove_wake_function+0x0/0x2d
  [<f890bc0f>] kjournald+0x0/0x184 [jbd]
  [<c0122b9c>] kthread+0xaf/0xdb
  [<c0122aed>] kthread+0x0/0xdb
  [<c0101005>] kernel_thread_helper+0x5/0xb
Code: 24 58 8b 40 20 89 44 24 58 39 f0 75 b3 8b 95 e8 00 00 00 89 54 24 
20 c7 44 24 24 00 00 00 00 e8 89 cb 96 c7 e9 f8 00 00 00 8b 37 <ff> 46 
30 8b 06 a8 02 75 0a c7 44 24 28 00 00 00 00 eb 42 0f ba
EIP: [<f89089bd>] journal_commit_transaction+0x22d/0xc08 [jbd] SS:ESP 
0068:f7819f3c
}}}


'''Opération post-incident :'''

'''08/08/2008'''

 * testmem86+ (liveCD Ubuntu) --> 55 erreurs !!

'''10/08/2008'''

 * Changement des deux barrettes de RAM ECC
 * testmem86+ --> ok
 * Erreur fsck sur /dev/sdb1......

'''11/08/2008'''

 * Boot en single user
 * umount /dev/sdb1
 * fsck -r /dev/sdb1
 * Un fichier corrompu, puis réparé !!
 * Redémarrage nickel !