In Ubuntu, on s390x (only), we've been noticing occasional failures with CLITest::testStdinPipe in phpmyadmin-sql-parser version 4.6.1. This test runs the php binary six times. Out of five different CI runs:
A: data set #2 and #5 failed
B: data set #2 and #3 failed
C: data set #1 and #3 failed
D: data set #3 and #4 failed
E: data sets #0, #1, #2, #3, #4, #5 all failed
https://autopkgtest.ubuntu.com/packages/phpmyadmin-sql-parser/groovy/s390x
In the cases where the CI passed, we weren't doing anything special, it seems to have just randomly passed.
I unfortunately don't have access to the s390x machine so am unable to investigate this directly, just via the CI system. The php cli and test suite pass properly on my (non-s390x) hardware. I'm hoping you may have some insights to help us make the test more deterministic?
In Ubuntu, on s390x (only), we've been noticing occasional failures with CLITest::testStdinPipe in phpmyadmin-sql-parser version 4.6.1. This test runs the php binary six times. Out of five different CI runs:
https://autopkgtest.ubuntu.com/packages/phpmyadmin-sql-parser/groovy/s390x
In the cases where the CI passed, we weren't doing anything special, it seems to have just randomly passed.
I unfortunately don't have access to the s390x machine so am unable to investigate this directly, just via the CI system. The php cli and test suite pass properly on my (non-s390x) hardware. I'm hoping you may have some insights to help us make the test more deterministic?