[
https://issues.apache.org/jira/browse/KNOX-3388?focusedWorklogId=1031624&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-1031624
]
ASF GitHub Bot logged work on KNOX-3388:
----------------------------------------
Author: ASF GitHub Bot
Created on: 22/Jul/26 07:45
Start Date: 22/Jul/26 07:45
Worklog Time Spent: 10m
Work Description: smolnar82 opened a new pull request, #1316:
URL: https://github.com/apache/knox/pull/1316
[KNOX-3388](https://issues.apache.org/jira/browse/KNOX-3388) - Batch
credential-store alias creation in single-EKU test entrypoint
## What changes were proposed in this pull request?
`gateway-single-eku.sh` created its 5 credential-store aliases via 5
separate `knoxcli.sh create-alias` calls. Each is a cold JVM boot (~6-8s), so
~30-40s was spent before the gateway JVM started. CI's `sleep 30` (measured
from container
start) was consumed by alias creation, so the gateway bound 8443 only after
`pytest` connected → intermittent `Connection refused` in the single-EKU mTLS
suite.
Replaced the loop with a single batch `knoxcli.sh create-aliases --alias ...
--value ...` invocation (one JVM boot). Container-start → `Started gateway on
port 8443` dropped from ~35s to ~9s, leaving comfortable margin under the
existing wait.
## How was this patch tested?
- Recreated the single-EKU knox locally and timed startup: gateway bound
8443 in ~9s (was ~35s).
- Ran the single-EKU mTLS suite: `pytest test_single_eku_mtls.py` → 3 passed.
## Integration Tests
No test changes needed — this fixes flakiness in the existing
`test_single_eku_mtls.py` suite under `.github/workflows/tests`.
## UI changes
N/A
Issue Time Tracking
-------------------
Worklog Id: (was: 1031624)
Remaining Estimate: 0h
Time Spent: 10m
> Single-EKU test gateway startup is slow/flaky: batch credential-store alias
> creation
> ------------------------------------------------------------------------------------
>
> Key: KNOX-3388
> URL: https://issues.apache.org/jira/browse/KNOX-3388
> Project: Apache Knox
> Issue Type: Improvement
> Affects Versions: 3.0.0
> Reporter: Sandor Molnar
> Assignee: Sandor Molnar
> Priority: Major
> Fix For: 3.0.0
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> The single-EKU compose entrypoint (gateway-single-eku.sh) creates its 5
> credential-store aliases with one {{knoxcli.sh create-alias}} call each.
> Every call is a cold JVM boot (~6-8s), so ~30-40s elapses before the gateway
> JVM even starts. In CI the "sleep 30" wait starts at container start and is
> fully consumed by alias creation, so the gateway binds 8443 only after
> {{pytest}} connects -> intermittent "Connection refused" in the single-EKU
> mTLS suite.
> Fix: create all 5 aliases in a single {{knoxcli.sh create-aliases}}
> invocation.
> Container-start -> "Started gateway on port 8443" drops from ~35s to ~9s.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
