-
Notifications
You must be signed in to change notification settings - Fork 3.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
sql: incorrect splitting lookup span into family spans #76289
Comments
roachtest.tlp failed with artifacts on master @ d10188f400deafa2569b052cd3e9fd9713783ed9:
Same failure on other branches
|
roachtest.tlp failed with artifacts on master @ 84a0e500920e6d0b935d2007de256e42822e94c7:
Same failure on other branches
|
roachtest.tlp failed with artifacts on master @ 80f1c2ce09389f1d7e97376964d3f2a922405b1b:
Same failure on other branches
|
roachtest.tlp failed with artifacts on master @ 260be01361ed7264a870e530397ed3529cd014f7:
Same failure on other branches
|
roachtest.tlp failed with artifacts on master @ bbb473c8f304ac20fec51ff0a0d04e128383bcf6:
Same failure on other branches
|
First failure is reduced to:
Here is where the problem appears:
Somehow we're missing a row when reading of a UNIQUE secondary index, and my guess is that we have a problem with UPDATE. |
The bug is present on both 21.1.13 and 21.2.5. I no longer think the problem is with the UPDATE because querying for a single column returns the correct result (with the plan being the same). I'm currently thinking that we're incorrectly constructing Get requests for the index join (on 21.1 the same applies to Scan requests for the index join). |
Smaller repro:
The order of column families matters, the order of column in index matters, DECIMAL cannot be replaced with INT8, non-zero value for DECIMAL has to be inserted. |
I think I've narrowed it down. The INSERT statements performs a single put for the first column family (
However, later in the index join we're trying to read value of
|
roachtest.tlp failed with artifacts on master @ 9514f13d0190849eda12b39e272c278b82c269f4:
Same failure on other branches
|
Hm, I cannot reproduce the latest failure (i.e. both queries return the same result sets):
@mgartner do you have an idea of why a failure would be filed in this case? |
@yuzefovich Reminds me of the bizarre failures I saw here: #74675 (comment). I think it's a problem with the test apparatus, but I haven't been able to confirm that. |
roachtest.tlp failed with artifacts on master @ 432a383c1a7509762ce66a2b9ae7869d82da5f31:
Help
See: roachtest README
See: How To Investigate (internal)
Same failure on other branches
This test on roachdash | Improve this report!
The text was updated successfully, but these errors were encountered: