[LU-17458] lnet-selftest doesn't work with large NIDs

• Status: Open
• Project: Lustre
• Type: Bug
• Priority: Minor
• Reporter: Chris Horn
• Assignee: WC Triage
• Resolution: Unresolved
• Severity: 3

Description

[root@el8-mds1 lst-survey]# ./lst.sh -t 2601:8c1:c180:2000::cbde@tcp -f 2601:8c1:c180:2000::cbdf@tcp -m read
Discover server NIDs
Discover client NIDs
Start LST read - Mon Jan 22 17:06:20 CST 2024
LST_SESSION=16659
SESSION: lnet_session FEATURES: 1 TIMEOUT: 300 FORCE: No
Adding clients: 2601:8c1:c180:2000::cbdf@tcp
Invalid nid: 2601:8c1:c180:2000::cbdf@tcp
Ignore invalid id list 2601:8c1:c180:2000::cbdf@tcp
No nodes added successfully, deleting group clients
Group is deleted
Adding servers: 2601:8c1:c180:2000::cbde@tcp
Invalid nid: 2601:8c1:c180:2000::cbde@tcp
Ignore invalid id list 2601:8c1:c180:2000::cbde@tcp
No nodes added successfully, deleting group servers
Group is deleted
Test: --batch brw_read --concurrency 16 --from clients --to servers --distribute 1:1 brw read size=1m
Stat: --count 1 --delay 15 --bw --mbs clients servers
Reader error: 'Object not found' at 0
Reader error: 'Object not found' at 0
^[[B./lst.sh: line 409: 16773 Killed                  lst add_test "${test_opts[@]}"
Stop LST read - Mon Jan 22 17:06:47 CST 2024
[root@el8-mds1 lst-survey]#

Comments

[Chris Horn]

Note, the failed lst commands are not returning a non-zero exit status:

[root@el8-mds1 lst-survey]# lst add_group clients 2601:8c1:c180:2000::cbde@tcp
Invalid nid: 2601:8c1:c180:2000::cbde@tcp
Ignore invalid id list 2601:8c1:c180:2000::cbde@tcp
No nodes added successfully, deleting group clients
Group is deleted
[root@el8-mds1 lst-survey]# echo $?
0
[root@el8-mds1 lst-survey]#