This message was deleted.

An HTTP 500 Internal Server Error during a Stream Load indicates that the StarRocks Backend (BE) encountered an unhandled exception or a crash while processing your request. The "index out of bounds" error you mentioned is likely the internal server-side message. To resolve this, follow these troubleshooting steps: 1. Check the BE logs for the actual stack trace The Java stack trace you provided is from your application client, which only shows that it received a 500 error. The root cause (the "index out of bounds") will be logged on the StarRocks BE nodes. * Connect to the BE node that received the request (the

StarRocksInserter

usually targets a specific BE or is routed there by the FE). * Search for the error in

${STARROCKS_HOME}/log/be.INFO

or

be.WARNING

. * Look for keywords like

IndexOutOfBounds

,

vector out-of-bounds

, or

Segmentation fault

. 2. Check for known "Index out of bounds" bugs If you are on a recent version of StarRocks (e.g., 3.1.x through 3.4.x), there are a few known issues that trigger this specific error: * JSON Parsing: A bug exists where calling

substr

or hyper-extracting JSON paths (when the path is empty or all paths are skipped) can cause an out-of-bounds exception (#65260). * String Functions: The

trim()

function has been known to trigger vector out-of-bounds operations on empty slices or specific Unicode characters (#66484). * Schema Mismatch: If your

columns

header or

jsonpaths

doesn't match the actual structure of the payload, it can occasionally trigger an internal indexing error instead of a graceful "quality" error. 3. Inspect the ErrorURL Even with a 500 error, StarRocks sometimes returns a JSON response containing an

ErrorURL

. * If your application logs the full HTTP response body, look for a field named

"ErrorURL"

. * Open that URL in a browser or via

curl

. it will show you exactly which row failed and why (e.g., "Value count does not match column count"). 4. Verify Column and Separator Config If you are loading CSV: * Ensure your

column_separator

in the header matches the actual file. A mismatch often causes StarRocks to see the entire row as a single column, leading to indexing errors when it tries to map fields. * Check if your

columns

header has more/fewer columns than the actual data. If you are loading JSON: * Verify that

jsonpaths

are correct. * If using

strip_outer_array: true

, ensure the payload is a valid JSON array. Next Steps If you find a specific C++ stack trace in the

<http://be.INFO|be.INFO>

log, please share that. If the BE is crashing (restarting automatically), check the

be.out

file for crash signals. For issues requiring deep investigation, consider providing your StarRocks version and the table schema. References • faq/loading/Loading_faq.md • administration/management/BE_configuration.md • faq/loading/Stream_load_faq.md • administration/management/logs.md • faq/loading/Broker_load_faq.md