[Error Situation]
A medium sometimes disappears from slot information within DataProtector(DP) unexpectedly although it exists physically in Tape Library .
For instance , a medium was tried to be loaded from slot18 to Drive3 as first .
However, another medium was loaded from slot17 to Drive3 without showing any reason.
Actually the medium exists in slot18 .
After this backup session , the media information about slot18 was set to "empty".
[Normal] From: BMA@hostname "Drive4" Time: 2016/10/05 15:27:12
STARTING Media Agent "Drive4"
[Normal] From: BMA@hostname "Drive3" Time: 2016/10/05 15:27:12
STARTING Media Agent "Drive3"
[Normal] From: BMA@hostname "Drive4" Time: 2016/10/05 15:27:12
=> UMA@hostname@/dev/sg81
Loading medium from slot 5 to device /dev/tape1_04
[Normal] From: BMA@hostname "Drive3" Time: 2016/10/05 15:27:12
=> UMA@hostname@/dev/sg81
Loading medium from slot 18 to device /dev/tape1_03
[Normal] From: BMA@hostname "Drive3" Time: 2016/10/05 15:28:20
=> UMA@hostname@/dev/sg81
Loading medium from slot 17 to device /dev/tape1_03
[Cause]
SCTL_ReadElementStatus will be initiated before moving media to confirm statuses of Slots and Drives in Tape Library .
To move a medium from slot18 to Drive3 ,SCTL_ReadElementStatus needed to finish accordingly but it reached time-out (60 seconds) because Tape Library was busy to move a medium to Drive4 and could not respond to it .
As the result , slot18 was recognized as "empty" incorrectly .
[Workaround]
The error situation can be avoided by setting omnirc variable OB2RELSTATSLEEP=600 on MediaAgent system to expand the time-out .
[Fix]
The issue will be solved by DP9.09 (QCCR2A71305) .
Now DP will notice about the time-out of SCTL_ReadElementStatus and it will be retried if it reached the time-out .
omnirc variables OB2SPTSLEEP and OB2SPTRETRY can be used to configure specific values for retry and sleep.
(Default : 10 seconds sleep and 10 times retry)