Browser OS version: Windows 11 Pro 23H2 Build 22631.4037
Original install method (e.g. download page, yum, from source, etc.): Deployed on Elastic Cloud
Describe the bug: In case too many anomaly jobs in the group, the overall score column is not displayed and the user gets "An HTTP line is larger than 4096 bytes" error.
Steps to reproduce:
Add anomaly jobs with the same group. Depending on the length of job IDs, but I could reproduce this with more than 80+ jobs.
Go to the ML Overview page
Expected behavior:
ML UI will work normally regardless of the number of jobs.
Screenshots (if relevant):
Any additional context:
The culprit was the below request (extracted from HAR), which is for fetching "Overall score" column.
https://<my_kibana_uri>/internal/ml/anomaly_detectors/clone__clone__long_long_very_long_job_id_to_reproduce_issue___1,clone__clone__long_long_very_long_job_id_to_reproduce_issue___2,...(total 4046 chars)/results/overall_buckets
# Response
{
"statusCode": 400,
"error": "Bad Request",
"message": "[too_long_http_line_exception\n\tRoot causes:\n\t\ttoo_long_http_line_exception: An HTTP line is larger than 4096 bytes.]: An HTTP line is larger than 4096 bytes.",
"attributes": {
"body": {
"error": {
"root_cause": [
{
"type": "too_long_http_line_exception",
"reason": "An HTTP line is larger than 4096 bytes."
}
],
"type": "too_long_http_line_exception",
"reason": "An HTTP line is larger than 4096 bytes."
},
"status": 400
}
}
}
I think the possible update is modifying this API to envelope the job IDs in the request body, or split the request into multiples by the total length.
Kibana version: 8.14.3
Elasticsearch version: 8.14.3
Server OS version: Elastic Cloud
Browser version: Microsoft Edge 127.0.2651.98
Browser OS version: Windows 11 Pro 23H2 Build 22631.4037
Original install method (e.g. download page, yum, from source, etc.): Deployed on Elastic Cloud
Describe the bug: In case too many anomaly jobs in the group, the overall score column is not displayed and the user gets "An HTTP line is larger than 4096 bytes" error.
Steps to reproduce:
Expected behavior: ML UI will work normally regardless of the number of jobs.
Screenshots (if relevant):
Any additional context: The culprit was the below request (extracted from HAR), which is for fetching "Overall score" column.
I think the possible update is modifying this API to envelope the job IDs in the request body, or split the request into multiples by the total length.