[
https://issues.apache.org/jira/browse/ARTEMIS-4197?focusedWorklogId=995493&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-995493
]
ASF GitHub Bot logged work on ARTEMIS-4197:
-------------------------------------------
Author: ASF GitHub Bot
Created on: 10/Dec/25 05:03
Start Date: 10/Dec/25 05:03
Worklog Time Spent: 10m
Work Description: jbertram commented on PR #6080:
URL: https://github.com/apache/artemis/pull/6080#issuecomment-3635397144
The full test-suite is green on this.
Issue Time Tracking
-------------------
Worklog Id: (was: 995493)
Time Spent: 20m (was: 10m)
> AbstractJournalStorageManager address-settings not thread-safe
> --------------------------------------------------------------
>
> Key: ARTEMIS-4197
> URL: https://issues.apache.org/jira/browse/ARTEMIS-4197
> Project: Artemis
> Issue Type: Bug
> Reporter: Jelmer Marinus
> Assignee: Justin Bertram
> Priority: Major
> Labels: pull-request-available
> Time Spent: 20m
> Remaining Estimate: 0h
>
> The AbstractJournalStorageManager contains the following code.
> {code:java}
> @Override
> public void storeAddressSetting(PersistedAddressSetting addressSetting)
> throws Exception {
> deleteAddressSetting(addressSetting.getAddressMatch());
> try (ArtemisCloseable lock = closeableReadLock()) {
> long id = idGenerator.generateID();
> addressSetting.setStoreId(id);
> bindingsJournal.appendAddRecord(id,
> JournalRecordIds.ADDRESS_SETTING_RECORD, addressSetting, true);
> mapPersistedAddressSettings.put(addressSetting.getAddressMatch(),
> addressSetting);
> }
> }{code}
> This code is used to update the AddressSettings through, i.e., the Artemis
> management API.
> The AbstractJournalStorageManager also contains the following codeĀ
> {code:java}
> @Override
> public List<PersistedAddressSetting> recoverAddressSettings() throws
> Exception {
> try (ArtemisCloseable lock = closeableReadLock()) {
> return new ArrayList<>(mapPersistedAddressSettings.values());
> }
> } {code}
> This code is called during, i.e., a config. reload when i.e. the broker.xml
> gets changed.
> When a first thread executes the first code fragement and a second thread
> executes the second fragment there is a possibility of the following:
> # The first thread executes the deleteAddressSettings for address MY_ADDRESS;
> # The second thread runs the whole recoverAddressSettings methodĀ
> # The first thread runs the remainder of the storeAddressSetting-method for
> MY_ADDRESS.
> According to the second thread there where no addressSettings for the address
> MY_ADDRESS. Because it ran in between of the delete and the put of the
> storeAddressSetting-method.
> I think this can be fixed using read-/write-locks from the storageManagerLock.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]