Siyao Meng created HDDS-16691:
---------------------------------
Summary: Retriable snapshot load TIMEOUT reaches clients as an
unchecked IllegalStateException
Key: HDDS-16691
URL: https://issues.apache.org/jira/browse/HDDS-16691
Project: Apache Ozone
Issue Type: Bug
Reporter: Siyao Meng
Attachments: HDDS-16691-CR-R11-load-timeout-unchecked.patch
h3. Mechanism
{{SnapshotCache.get}} tolerates only {{FILE_NOT_FOUND}} and rewraps every other
{{OMException}} as {{IllegalStateException}} ({{SnapshotCache.java:230-234}}).
The loader raises {{OMException(TIMEOUT)}} when a snapshot's create transaction
has not yet been flushed ({{OmSnapshotManager.java:380-383}}), which is the
very condition the HDDS-13633 poll exists for.
The wrapping exists because the lambda passed to {{dbMap.compute}} cannot throw
checked exceptions. {{FILE_NOT_FOUND}} was given an escape hatch and
{{TIMEOUT}} was not.
h3. Impact
A retriable snapshot load TIMEOUT reaches the read request handler as an
unchecked {{IllegalStateException}} it cannot map, so the client gets an
unstructured {{RemoteException}} instead of status {{TIMEOUT}} and loses the
retry hint. Client visible on every occurrence, bounded, and not persistent in
server state.
h3. Reproduction
{{repro/test_bugCR-R11_timeout_becomes_unchecked.sh}}. Evidence in
{{repro/logs/ev-cr2.log}}.
h3. Patch
{{HDDS-XXXXX-CR-R11-load-timeout-unchecked.patch}} is attached. It applies to
20e6a1c0f6039248373efd3eb1d7f2afcf7f1535 and carries the fix plus a regression
test that was verified to fail without the fix and pass with it.
{code:none}
mvn test -pl hadoop-ozone/ozone-manager -Dtest=TestSnapshotCache
{code}
Found by TLA+ model checking and code review of the OM snapshot cache under
HDDS-15926, on commit 20e6a1c0f6039248373efd3eb1d7f2afcf7f1535. Generated with
Specula (Claude Opus 5).
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
