> On Oct 2, 2020, at 1:42 PM, J. Bruce Fields <bfields@xxxxxxxxxxxx> wrote:
>
> On Fri, Oct 02, 2020 at 01:39:08PM -0400, J. Bruce Fields wrote:
>> I'm seeing a pynfs4.0 GATT9 regression. That's a test that attempts a
>> compound with 90 GETATTR ops each a request for all mandatory
>> attributes. The test expects OK or RESOURCE but looks like its getting
>> a corrupted response? (I haven't looked at the wire traffic yet.) I
>> think it's one of the final patches changing how errors are returned.
>
> Also some other tests that send compounds with lots of ops:
>
> GATT9 st_getattr.testLotsofGetattrsFile : FAILURE
> nfs4lib.InvalidCompoundRes: Invalid COMPOUND result:
> Truncated response list.
> COMP6 st_compound.testLongCompound : FAILURE
> COMPOUND with len=150 argarry got Invalid COMPOUND
> result: Truncated response list., expected
> NFS4ERR_RESOURCE
> COMP4 st_compound.testInvalidMinor : FAILURE
> nfs4lib.InvalidCompoundRes: Invalid COMPOUND result:
> Truncated response list.
>
> Bisect lands on the last patch ("Hoist status code...").
Excellent, thanks for the test results. I'll take a look.
--
Chuck Lever
