In the operation of Linux servers using Emulex Fiber card database server to store large probability of a link failure herein, this only describes the optical card replacement program, storage link failures do follow-up detail.
/ Var / log / message given information is as follows:
kernel: lpfc 0000:07:00.0: 0:(0):0748 abort handler timed out waiting for abort to complete: ret 0x2003, ID 3, LUN 4, snum 0xa09e9c0
kernel: INFO: task kmirrord:7801 blocked for more than 120 seconds.
kernel: "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
Replacing the optical card steps:
1, stop the service and shut down, to note here have mounted storage system boot time, namely / etc / fstab have storage paths are commented out before shutting down, otherwise the boot time will not start, such as forgetting Notes can not start the Notes required to enter maintenance mode mount -o remount, rw /, edit / etc / fstab;
2, replace the Qlogic fiber card, I am here to replace all the Qlogic Emulex after the problem;
3, the storage side delete the old link to be removed in the Storage Group host can be added. I used here is the EMC VNX series storage, optical cross-end do not need to operate;
4, the boot server, Scan link, echo "- - -"> / sys / class / scsi_host / host # / scan or echo "1"> / sys / class / fc_host / host / issue_lip, and then view the status of the link, powerpath use powermt display dev = all, multipath using multipath -ll, can also perform pvs, vgs, lvs check the volume is normal; such as the error message "could not find device with uuid XXXX" appear, to note here is not a disk the UUID changed, but even after replacing the optical card by scanning all of the disks have not fully found, then you can re-enable and disable scan repaired by fiber card, or restart the server directly recovery;