unit test failures

57 views
Skip to first unread message

ravisa...@gmail.com

unread,
Apr 30, 2019, 5:37:47 AM4/30/19
to CockroachDB
I am working on s390x by building it from source. 
I have built cockroach-db v2.6.1.

Following tests got failed:

Log using TESTFLAGS=-v are in the attachement .

Any input on this will be very useful.


Regards,
Ravi Salamani




make_test_pkg_sql_pgwire.txt
make_test_pkg_ts.txt
make_test_sql_sqllogic_cmd.txt

Radu Berinde

unread,
Apr 30, 2019, 9:00:20 AM4/30/19
to ravisa...@gmail.com, CockroachDB
Hi Ravi,

I assume you meant v2.1.6.  Note that 19.1 is about to be released, you might as well switch to building from release-19.1.

The logictest is failing because of a minor difference (last digit) in a floating point result; we should make the test more tolerant (perhaps by rounding to fewer digits)

The ts failure is likely due to differences in bit / arithmetic ops, and is worth investigating. I'm not sure how we would get access to a s390x system though.

The pgwire failure is fixed by https://github.com/cockroachdb/cockroach/pull/36212, if you try to build the latest on release-2.1 or release-19.1 it should be gone.

-Radu


--
You received this message because you are subscribed to the Google Groups "CockroachDB" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cockroach-db...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cockroach-db/ad1660b9-d95f-4d28-b758-ae888c51610a%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

ravisa...@gmail.com

unread,
May 3, 2019, 9:23:32 AM5/3/19
to CockroachDB
Hi,

Thanks radu for your input. 
I have build the release-19.1 now.
pgwire failure fixed. remaining are same.
I found new failure github.com/cockroachdb/cockroach/pkg/sql/opt/xform, Plz find the attachment for log. Any idea on this?
 

Regards,
Ravi Salamani

On Tuesday, April 30, 2019 at 6:30:20 PM UTC+5:30, Radu Berinde wrote:
Hi Ravi,

I assume you meant v2.1.6.  Note that 19.1 is about to be released, you might as well switch to building from release-19.1.

The logictest is failing because of a minor difference (last digit) in a floating point result; we should make the test more tolerant (perhaps by rounding to fewer digits)

The ts failure is likely due to differences in bit / arithmetic ops, and is worth investigating. I'm not sure how we would get access to a s390x system though.

The pgwire failure is fixed by https://github.com/cockroachdb/cockroach/pull/36212, if you try to build the latest on release-2.1 or release-19.1 it should be gone.

-Radu


On Tue, Apr 30, 2019 at 5:37 AM <ravisa...@gmail.com> wrote:
I am working on s390x by building it from source. 
I have built cockroach-db v2.6.1.

Following tests got failed:

Log using TESTFLAGS=-v are in the attachement .

Any input on this will be very useful.


Regards,
Ravi Salamani




--
You received this message because you are subscribed to the Google Groups "CockroachDB" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cockro...@googlegroups.com.

ravisa...@gmail.com

unread,
May 3, 2019, 9:25:04 AM5/3/19
to CockroachDB
Adding attachment for log
sql_opt_xform_log.txt

ravisa...@gmail.com

unread,
May 6, 2019, 6:24:38 AM5/6/19
to CockroachDB

Hi All,
i.e.,
--- FAIL: TestExternal (26.67s)
    --- FAIL: TestExternal/tpcc (0.02s)
        optimizer_test.go:229:
Plz find the logs in old mail.
Failure is likely due to minor difference in float value right?

Regards,
Ravi Salamani

Radu Berinde

unread,
May 6, 2019, 9:06:13 AM5/6/19
to ravisa...@gmail.com, CockroachDB
Yeah, just a minor difference in the internal costs. We could probably fix by displaying fewer digits.

Please file issues for these failures, it's easier to keep track of them that way.

-Radu


To unsubscribe from this group and stop receiving emails from it, send an email to cockroach-db...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cockroach-db/2d9d5cb5-c4d8-47ba-9b8a-4b644a1ff722%40googlegroups.com.

ravisa...@gmail.com

unread,
May 7, 2019, 2:15:12 AM5/7/19
to CockroachDB
sure. I will raise a issue.
I can see ts failure is also due single digit difference. Is it also the same case?

$ make test PKG=./pkg/ts TESTS='TestTimeSeriesRollupModelTest$$' TESTFLAGS="-v -show-logs"
```
Running make with -j2
GOPATH set to /home/ubuntu/cockroach_work
=== RUN   TestTimeSeriesRollupModelTest
I190507 06:01:08.397706 27 gossip/gossip.go:392  [n1] NodeDescriptor set to node_id:1 address:<network_field:"tcp" address_field:"invalid.invalid:26257" > attrs:<> locality:<> ServerVersion:<major_val:0 minor_val:0 patch:0 unstable:0 > build_tag:"" started_at:0
W190507 06:01:08.397969 65 storage/store.go:1525  [n1,s1,r6/1:/Table/{SystemConâ¦-11}] could not gossip system config: [NotLeaseHolderError] r6: replica (n1,s1):1 not lease holder; lease holder unknown
--- FAIL: TestTimeSeriesRollupModelTest (2.32s)
    model_test.go:249: ["\x04tsd\x12metric1\x00\x01\x8a\xf7D{source4"].RawBytes[80]: 199 != 198
    model_test.go:249: ["\x04tsd\x12metric2\x00\x01\x8a\xf7D{source4"].RawBytes[80]: 161 != 160
    model_test.go:249: ["\x04tsd\x12metric1\x00\x01\x8a\xf7D{source1"].RawBytes[80]: 51 != 54
    model_test.go:249: ["\x04tsd\x12metric1\x00\x01\x8a\xf7D{source3"].RawBytes[80]: 198 != 197
    model_test.go:249: ["\x04tsd\x12metric2\x00\x01\x8a\xf7D{source1"].RawBytes[80]: 89 != 90
FAIL
Makefile:893: recipe for target 'test' failed
make: *** [test] Error 1
```

Regards,
Ravi Salamani

Radu Berinde

unread,
May 7, 2019, 7:36:41 AM5/7/19
to ravisa...@gmail.com, CockroachDB
Yes, it's probably the same underlying issue (floating point difference).

-Radu


To unsubscribe from this group and stop receiving emails from it, send an email to cockroach-db...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cockroach-db/b6f22e0d-e453-4d8f-a9ba-20bb1f948304%40googlegroups.com.

Ravi Salamani

unread,
May 13, 2019, 1:24:46 AM5/13/19
to CockroachDB
I have created issue #37423 with all issues faced for s390x to keep track of it. 

Regards,
Ravi Salamani
Reply all
Reply to author
Forward
0 new messages