[LU-11728] parallel-scale test connectathon fails with ''connectathon failed: 1''

• Type: Bug
• Resolution: Unresolved
• Priority: Minor
• Affects Version/s: Lustre 2.10.6
• Labels: DNE zfs
• Environment: DNE/ZFS
• Severity: 3
• Assignee: WC Triage
• Reporter: James Nunez (Inactive)

Description

parallel-scale test connectathon fails with ''connectathon failed: 1''

Looking at the logs at https://testing.whamcloud.com/test_sets/bc3cdec6-f6df-11e8-86c0-52540065bddc, the final lines of parallel-scale test_connectathon are

test rewind support

test telldir cookies
expected file 90 at cookie 0, found .
special tests failed
 parallel-scale test_connectathon: @@@@@@ FAIL: connectathon failed: 1 

There are no errors in the dmesg nor console logs.

This failure looks like LU-9322, but there are no errors in the console logs as in LU-9322.

This is the first time this error has been seen in at least the past seven months. So far, this has only failed for ZFS/DNE configurations.

Issue Links

• is related to: LU-9322 parallel-scale test_connectathon: connectathon failed: 1

Comments

[James Nunez (Inactive)]

I don't see any evidence that the connectathon test was run with a striped directory.

[Andreas Dilger]

The failing test "telldir cookies" relates to being able to restart "readdir()" operations at a specific spot. It isn't totally surprising that this would be having problems on a DNE striped directory (if that is what is being tested), since this is one of the more complex parts of DNE to combine. However, it isn't totally clear to me what the test is trying to do. The Lustre logs show a lot of seeking around in the directory, and then finally the test fails when it returns to directory offset 0. There are a large number of connectathon tests passing, so it isn't necessarily a critical/common issue.