This is an automated email from the ASF dual-hosted git repository.
yiguolei pushed a commit to branch branch-4.1
in repository https://gitbox.apache.org/repos/asf/doris.git
The following commit(s) were added to refs/heads/branch-4.1 by this push:
new 04c3a97a912 branch-4.1: [fix](arrow-flight) Harden split source error
path to avoid BE crash on external table scan #64797 (#64890)
04c3a97a912 is described below
commit 04c3a97a9127ef28e3ae458c6a869d8c970f9879
Author: github-actions[bot]
<41898282+github-actions[bot]@users.noreply.github.com>
AuthorDate: Fri Jun 26 20:50:01 2026 +0800
branch-4.1: [fix](arrow-flight) Harden split source error path to avoid BE
crash on external table scan #64797 (#64890)
Cherry-picked from #64797
Co-authored-by: Mingyu Chen (Rayner) <[email protected]>
Co-authored-by: Claude Opus 4.8 (1M context) <[email protected]>
---
be/src/exec/scan/split_source_connector.cpp | 4 +++-
be/src/format/arrow/arrow_utils.cpp | 12 +++++++++++-
.../java/org/apache/doris/service/FrontendServiceImpl.java | 10 +++++++++-
3 files changed, 23 insertions(+), 3 deletions(-)
diff --git a/be/src/exec/scan/split_source_connector.cpp
b/be/src/exec/scan/split_source_connector.cpp
index 81748261d98..685a2d50f0c 100644
--- a/be/src/exec/scan/split_source_connector.cpp
+++ b/be/src/exec/scan/split_source_connector.cpp
@@ -59,7 +59,9 @@ Status RemoteSplitSourceConnector::get_next(bool* has_next,
TFileRangeDesc* rang
coord->fetchSplitBatch(result, request);
if (result.__isset.status && result.status.status_code !=
TStatusCode::OK) {
return Status::IOError<false>("Failed to get batch of split
source: {}",
- result.status.error_msgs[0]);
+ result.status.error_msgs.empty()
+ ? "unknown error"
+ :
result.status.error_msgs[0]);
}
} catch (std::exception& e) {
return Status::IOError<false>("Failed to get batch of split
source: {}", e.what());
diff --git a/be/src/format/arrow/arrow_utils.cpp
b/be/src/format/arrow/arrow_utils.cpp
index 724af28be39..d9c29eceed0 100644
--- a/be/src/format/arrow/arrow_utils.cpp
+++ b/be/src/format/arrow/arrow_utils.cpp
@@ -38,7 +38,17 @@ arrow::Status to_arrow_status(const Status& status) {
// The length of exception msg returned to the ADBC Client cannot
larger than 8192,
// otherwise ADBC Client will receive:
// `INTERNAL: http2 exception Header size exceeded max allowed size
(8192)`.
- return arrow::Status::Invalid(status.to_string_no_stack());
+ // The message is carried in the gRPC trailer (an HTTP2 header) and
may be
+ // percent-encoded (which can expand its size), so an oversized
message can break the
+ // response or even crash the flight transport status conversion.
Truncate it well below
+ // 8192 to leave headroom; the full message is already logged above.
+ constexpr size_t kMaxArrowStatusMsgLen = 1024;
+ std::string msg = status.to_string_no_stack();
+ if (msg.size() > kMaxArrowStatusMsgLen) {
+ msg.resize(kMaxArrowStatusMsgLen);
+ msg += "... (truncated)";
+ }
+ return arrow::Status::Invalid(msg);
}
}
diff --git
a/fe/fe-core/src/main/java/org/apache/doris/service/FrontendServiceImpl.java
b/fe/fe-core/src/main/java/org/apache/doris/service/FrontendServiceImpl.java
index 7441113ad56..f0d70434183 100644
--- a/fe/fe-core/src/main/java/org/apache/doris/service/FrontendServiceImpl.java
+++ b/fe/fe-core/src/main/java/org/apache/doris/service/FrontendServiceImpl.java
@@ -1023,7 +1023,15 @@ public class FrontendServiceImpl implements
FrontendService.Iface {
SplitSource splitSource =
Env.getCurrentEnv().getSplitSourceManager().getSplitSource(request.getSplitSourceId());
if (splitSource == null) {
- throw new TException("Split source " + request.getSplitSourceId()
+ " is released");
+ // Return a structured error status instead of throwing a bare
TException, so the BE
+ // receives a well-formed error (with a non-empty message) through
the normal result
+ // path. Throwing here surfaces as a thrift transport exception on
the BE side and,
+ // on the Arrow Flight path, may feed an empty/invalid error
string into the gRPC
+ // status conversion. See
https://github.com/apache/doris/issues/62259
+ LOG.warn("split source {} is released",
request.getSplitSourceId());
+ result.status = new TStatus(TStatusCode.NOT_FOUND);
+ result.status.addToErrorMsgs("Split source " +
request.getSplitSourceId() + " is released");
+ return result;
}
try {
List<TScanRangeLocations> locations =
splitSource.getNextBatch(request.getMaxNumSplits());
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
