Getting repeated 'Failed to get completion for aborted cmd xxxx' errors in logs and filesystem is switched to Read-Only mode
Issue
-
While creating a snapshot of virtual machine using
Veeam
utility, IO requests on VMware virtual disks are taking too long to complete and logs are gets filled with following errors:scsi_eh_2: aborting cmd:0xffff88022dfad7c0 sd 2:0:0:0: [sda] task abort on host 2, ffff88022dfad7c0 sd 2:0:0:0: [sda] Failed to get completion for aborted cmd ffff88022dfad7c0 scsi_eh_2: aborting cmd failed:0xffff88022dfad7c0 scsi_eh_2: aborting cmd:0xffff88022dfad2c0 sd 2:0:0:0: [sda] task abort on host 2, ffff88022dfad2c0 sd 2:0:0:0: [sda] Failed to get completion for aborted cmd ffff88022dfad2c0 scsi_eh_2: aborting cmd failed:0xffff88022dfad2c0
-
In some of the instances, root filesystem which is present on virtual disks remounts in read-only mode after above errors
kernel: end_request: I/O error, dev sda, sector 24640208 kernel: Aborting journal on device dm-4-8. kernel: EXT4-fs error (device dm-4) in ext4_reserve_inode_write: Journal has aborted kernel: EXT4-fs error (device dm-4) in ext4_dirty_inode: Journal has aborted kernel: EXT4-fs error (device dm-4): ext4_journal_start_sb: Detected aborted journal kernel: EXT4-fs (dm-4): Remounting filesystem read-only
-
Same issue still persists after increasing SCSI timeout for VMware virtual disks to 180 seconds as suggested in solution 35329
Environment
- Red Hat Enterprise Linux 6, 7 as a VMware guest system
- VMware virtual disks connected through PVSCSI adapter
Veeam
snapshot, disaster-recovery utility
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.