This looks caused by
scsi: Don't free IO channel at LUN removal not by allocator ·
IO channel is freed unexpectedly at LUN removal. This hides error due to uncorrect
implementation of LUN hot removal. IO channel of LUN must be freed only by its allocator.
And I should have done the following also for ext4test (probably more)
Currently re-enabling iSCSI hot plug is going forward.
Hence I want to and will change the ordering of ext4test.
I will investigate and think about it carefully.
差出人: SPDK <spdk-bounces(a)lists.01.org> が Verkamp, Daniel
送信日時: 2018年6月30日 2:05
宛先: Storage Performance Development Kit
件名: [!][SPDK] Nightly iSCSI test failing
It appears that test/iscsi_tgt/ext4test/ext4test.sh has been failing on master since the
06/27/18 nightly test run.
Last passing build:
Based on master commit 8c71c208a2c478dbd2a8dd5a4815c6591be5d670
First failing build:
Based on master commit 322e50468cb79224697b01a0e87ce9f24b012a38
Commits in this range:
The problem seems to be related to changes in iSCSI causing the delete_error_bdev RPC
method to hang (eventually timing out).
Could someone please investigate and follow up in this thread?
SPDK mailing list