[
https://issues.apache.org/jira/browse/HDDS-8268?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Siyao Meng updated HDDS-8268:
-----------------------------
Description:
We could try to enable some [spotbugs|https://spotbugs.github.io/] that could
have caught a bunch of not-closed handles before a commit is merged.
For instance,
bq. OBL: Method may fail to clean up stream or resource
(OBL_UNSATISFIED_OBLIGATION)
https://spotbugs.readthedocs.io/en/stable/bugDescriptions.html#obl-method-may-fail-to-clean-up-stream-or-resource-obl-unsatisfied-obligation
and
bq. OBL: Method may fail to clean up stream or resource on checked exception
(OBL_UNSATISFIED_OBLIGATION_EXCEPTION_EDGE)
https://spotbugs.readthedocs.io/en/stable/bugDescriptions.html#obl-method-may-fail-to-clean-up-stream-or-resource-on-checked-exception-obl-unsatisfied-obligation-exception-edge
could come in handy.
Just recently we have quite a few PRs just to fix the leaking handles:
1. HDDS-8251. RocksIterator is not closed properly in snapshot
2. HDDS-8265. ManagedColumnFamilyOptions is not closed properly in
OmSnapshotManager
3. HDDS-8163. Use try-with-resources to ensure close rockdb connection in
SstFilteringService
4. HDDS-7958. Ozone client not closed in integration tests
5. HDDS-8145. ReadReplicas should close client
Not sure about the false-positive rate though. Have to try to find out.
---
Looks like we explicitly disabled the checks for some of the tests:
https://github.com/apache/ozone/blob/68eb26d70e43648ea9270d9efa7664312a1947f9/hadoop-ozone/tools/dev-support/findbugsExcludeFile.xml#L17-L20
https://github.com/apache/ozone/blob/68eb26d70e43648ea9270d9efa7664312a1947f9/hadoop-ozone/recon/dev-support/findbugsExcludeFile.xml#L42-L53
Does it mean the handle leak checks are already enabled for the rest?
was:
We could try to enable some [spotbugs|https://spotbugs.github.io/] that could
have caught a bunch of not-closed handles before a commit is merged.
For instance,
bq. OBL: Method may fail to clean up stream or resource
(OBL_UNSATISFIED_OBLIGATION)
https://spotbugs.readthedocs.io/en/stable/bugDescriptions.html#obl-method-may-fail-to-clean-up-stream-or-resource-obl-unsatisfied-obligation
and
bq. OBL: Method may fail to clean up stream or resource on checked exception
(OBL_UNSATISFIED_OBLIGATION_EXCEPTION_EDGE)
https://spotbugs.readthedocs.io/en/stable/bugDescriptions.html#obl-method-may-fail-to-clean-up-stream-or-resource-on-checked-exception-obl-unsatisfied-obligation-exception-edge
could come in handy.
Just recently we have quite a few PRs just to fix the leaking handles:
1. HDDS-8251. RocksIterator is not closed properly in snapshot
2. HDDS-8265. ManagedColumnFamilyOptions is not closed properly in
OmSnapshotManager
3. HDDS-8163. Use try-with-resources to ensure close rockdb connection in
SstFilteringService
4. HDDS-7958. Ozone client not closed in integration tests
5. HDDS-8145. ReadReplicas should close client
Not sure about the false-positive rate though. Have to try to find out.
---
Looks like we explicitly disabled the checks for some of the tests:
https://github.com/apache/ozone/blob/68eb26d70e43648ea9270d9efa7664312a1947f9/hadoop-ozone/tools/dev-support/findbugsExcludeFile.xml#L17-L20
https://github.com/apache/ozone/blob/68eb26d70e43648ea9270d9efa7664312a1947f9/hadoop-ozone/recon/dev-support/findbugsExcludeFile.xml#L42-L53
Does it mean they are enabled already for the rest?
> spotbugs: enable resource clean up check
> ----------------------------------------
>
> Key: HDDS-8268
> URL: https://issues.apache.org/jira/browse/HDDS-8268
> Project: Apache Ozone
> Issue Type: Wish
> Components: CI
> Reporter: Siyao Meng
> Priority: Major
>
> We could try to enable some [spotbugs|https://spotbugs.github.io/] that could
> have caught a bunch of not-closed handles before a commit is merged.
> For instance,
> bq. OBL: Method may fail to clean up stream or resource
> (OBL_UNSATISFIED_OBLIGATION)
> https://spotbugs.readthedocs.io/en/stable/bugDescriptions.html#obl-method-may-fail-to-clean-up-stream-or-resource-obl-unsatisfied-obligation
> and
> bq. OBL: Method may fail to clean up stream or resource on checked exception
> (OBL_UNSATISFIED_OBLIGATION_EXCEPTION_EDGE)
> https://spotbugs.readthedocs.io/en/stable/bugDescriptions.html#obl-method-may-fail-to-clean-up-stream-or-resource-on-checked-exception-obl-unsatisfied-obligation-exception-edge
> could come in handy.
> Just recently we have quite a few PRs just to fix the leaking handles:
> 1. HDDS-8251. RocksIterator is not closed properly in snapshot
> 2. HDDS-8265. ManagedColumnFamilyOptions is not closed properly in
> OmSnapshotManager
> 3. HDDS-8163. Use try-with-resources to ensure close rockdb connection in
> SstFilteringService
> 4. HDDS-7958. Ozone client not closed in integration tests
> 5. HDDS-8145. ReadReplicas should close client
> Not sure about the false-positive rate though. Have to try to find out.
> ---
> Looks like we explicitly disabled the checks for some of the tests:
> https://github.com/apache/ozone/blob/68eb26d70e43648ea9270d9efa7664312a1947f9/hadoop-ozone/tools/dev-support/findbugsExcludeFile.xml#L17-L20
> https://github.com/apache/ozone/blob/68eb26d70e43648ea9270d9efa7664312a1947f9/hadoop-ozone/recon/dev-support/findbugsExcludeFile.xml#L42-L53
> Does it mean the handle leak checks are already enabled for the rest?
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
