 |
Maybe adjusting the mtx-changer.conf value of # Set to amount of time in seconds to wait after a load load_sleep=0 and also may it is needed to hack the mtx-changer script itself to add more time while [ $i -le 300 ]; do # Wait max 300 seconds |
|
|
I decided against changing "load_sleep" as this would affect all tape loads, but a longer timeout is only needed on the first load. If every tape load would have a delay of 2 hours the backup process would be very tedious. I modified the wait_for_drive() function in the mtx-changer script instead: wait_for_drive() { i=0 while [ $i -le 8000 ]; do # Wait max 2.22 hours - LTO-9 tapes need 2 hours calibration on first load debug "Doing mt -f $1 status ..." drivestatus=$(mt -f "$1" status 2>&1) if echo "${drivestatus}" | grep "${ready}" >/dev/null 2>&1; then break fi debug "${drivestatus}" debug "Device $1 - not ready, retrying ..." sleep 100 #was 'sleep 1' - do not poll the drive so often i=`expr $i + 100` done } but this didn't work. I ran into the same error: Sending label command for Volume "NSL142L9" Slot 13 ... 3304 Issuing autochanger "load slot 13, drive 0" command. 3992 Bad autochanger "load slot 13, drive 0": ERR=Child died from signal 15: Termination. Results=Program killed by BAREOS (timeout) Label command failed for Volume NSL142L9. The problem behind this is that the wait inside the wait_for_drive() function has no effect as the call to "mt -f /dev/nsa0 status" does not return at all while the tape drive is calibrating the LTO-9 tape. So even the original wait of 300 seconds would not have elapsed as the call of "mt -f /dev/nsa0 status" never returned. There seems to be another timeout somewhere kicking in that kills the label command. |
|
|
We will normally introduce a new parameter for this timeout see the proposal in https://github.com/bareos/bareos/pull/1740 You may want to test directly the script proposed in the PR |
|
|
a new parameter max_wait_drive=300 is proposed in the configuration file and used by the script. |
- Anonymous
