Hi Devs, Lately we see permanent CI issues [1] on many branches, like master, 2.0, 2.1, 2.2. There is jira also created for it [2]. Since we're intended to release 2.0.1 and 2.1.1 it would be good to see green azure runs for pending PRs.
I've just had an in-depth look for the issue and all I think it’s not a Flink problem - the Azure agent crashes. Right after the Docker image is built, a Go binary on the agent segfaults in Microsoft’s FIPS-enabled Go OpenSSL shim: *panic: runtime error: invalid memory address or nil pointer dereference vendor/github.com/golang-fips/openssl/v2.openLibrary.func2() <http://github.com/golang-fips/openssl/v2.openLibrary.func2()> … /usr/lib/msft-golang/.../opensslsetup.go (SIGSEGV) dev.azure.com <http://dev.azure.com>* Because of that crash, the “kubernetes session” e2e test aborts: *[FAIL] 'Run kubernetes session test (default input)' failed … Test exited with exit code 1 * So the pipeline is failing due to the agent’s Go/OpenSSL (FIPS) runtime blowing up—likely a mismatch with the OpenSSL library on the Ubuntu image - rather than a Flink test/assertion failure. Since I'm far away from any azure pipeline knowledge I would appreciate any help. BR, G [1] https://dev.azure.com/apache-flink/98463496-1af2-4620-8eab-a2ecc1a2e6fe/_build/results?buildId=70100 [2] https://issues.apache.org/jira/browse/FLINK-38487
