[LU-13048] Non-blocking migrate: Unable to release file from disk (Cannot send HSM request)

• Type: Bug
• Resolution: Fixed
• Priority: Major
• Fix Version/s: Lustre 2.16.0
• Severity: 3
• Assignee: Guillaume Courrier
• Reporter: Artem Blagodarenko (Inactive)

Description

The same root-case like in LU-5896, but about non-blocking migrate.
Non blocking swap layout (lfs migrate -n) uses different request and need separate fix.
old and new object version need to be added to struct close_data.
To reproduce add -n to migrate operation in sanity-ham 607
== sanity-hsm test 607: release migrated archived file =============================================== 03:24:42 (1574843082)
1+0 records in
1+0 records out
1048576 bytes (1.0 MB) copied, 0.0379826 s, 27.6 MB/s
Starting copytool agt1 on localhost.localdomain
Waiting 200 secs for update
Cannot send HSM request (use of /mnt/lustre/d607.sanity-hsm/foodir/f607.sanity-hsm): Device or resource busy
sanity-hsm test_607: @@@@@@ FAIL: cannot release /mnt/lustre/d607.sanity-hsm/foodir/f607.sanity-hsm
  Trace dump:
  = /lustre/mnt/orig/cray-lustre/lustre/tests/test-framework.sh:5988:error()
  = lustre/tests/sanity-hsm.sh:5952:test_607()
  = /lustre/mnt/orig/cray-lustre/lustre/tests/test-framework.sh:6272:run_one()
  = /lustre/mnt/orig/cray-lustre/lustre/tests/test-framework.sh:6311:run_one_logged()
  = /lustre/mnt/orig/cray-lustre/lustre/tests/test-framework.sh:6107:run_test()
  = lustre/tests/sanity-hsm.sh:5957:main()
Dumping lctl log to /tmp/test_logs/1574843058/sanity-hsm.test_607.*.1574843083.log
Dumping logs only on local client.
Resetting fail_loc on all nodes...done.
FAIL 607 (2s)
mdt.lustre-MDT0000.hsm_control=shutdown
== sanity-hsm test complete, duration 26 sec ========================================================= 03:24:44 (1574843084

Issue Links

• is related to: LU-5896 lfs: can't release a migrated file