Github user aljoscha commented on the issue:
https://github.com/apache/flink/pull/5896
@suez1224 Did you rebase on master? I think that krb5-specific could should
not be there anymore.
---
Github user aljoscha commented on the issue:
https://github.com/apache/flink/pull/5896
@EronWright the fix is really one this:
https://github.com/apache/flink/commit/ba3e2711a2091b7c0907a5caf3ea527f837a442f
But I haven't yet managed to reproduce a failure on my system without the fix.
Github user EronWright commented on the issue:
https://github.com/apache/flink/pull/5896
This PR seems to obfuscate the fix. The issue was with the interpretation
of the keytab path, right? But the bulk of the change was to clarify the
ordering of context installation vs use?
---
Github user suez1224 commented on the issue:
https://github.com/apache/flink/pull/5896
@aljoscha that's a very good catch. I added another commit to make sure the
call order is always correct
(https://github.com/apache/flink/pull/5896/commits/1298b04ea80b9160ac7e4ae8a171f80112a95945).
Github user aljoscha commented on the issue:
https://github.com/apache/flink/pull/5896
Merged, could you please close the PR after discussion of the refactoring?
---
Github user aljoscha commented on the issue:
https://github.com/apache/flink/pull/5896
@suez1224 I'll now merge the actual fix, but I'm not 100 % sure the
refactoring is correct.
After the fix, we have roughly this path through the code:
```
if (keytabPath != null && r