[ https://issues.apache.org/jira/browse/TEPHRA-270?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16453268#comment-16453268 ]
ASF GitHub Bot commented on TEPHRA-270: --------------------------------------- GitHub user poornachandra opened a pull request: https://github.com/apache/incubator-tephra/pull/72 TEPHRA-270 TEPHRA-271 Transaction state cache bug fixes JIRA: https://issues.apache.org/jira/browse/TEPHRA-270 https://issues.apache.org/jira/browse/TEPHRA-271 This PR contains a couple of bug fixes for the Transaction state cache to fix race conditions and deadlocks. You can merge this pull request into a Git repository by running: $ git pull https://github.com/poornachandra/incubator-tephra feature/tx-state-cache Alternatively you can review and apply these changes as the patch at: https://github.com/apache/incubator-tephra/pull/72.patch To close this pull request, make a commit to your master/trunk branch with (at least) the following in the commit message: This closes #72 ---- commit 73feed9a6a93d032c2e19e4e7d77fa4eb2065fcf Author: poorna <poorna@...> Date: 2018-04-25T23:51:56Z TEPHRA-270 Do not refresh the transaction state during co-processor startup to prevent deadlocks commit d42b282ec03c8e0d07bbb2a714961f2dd9b7641d Author: poorna <poorna@...> Date: 2018-04-25T23:58:22Z TEPHRA-271 Wait for the refresh thread to stop during shutdown to prevent race conditions ---- > Avoid transaction state cache refresh on co-procesor startup > ------------------------------------------------------------ > > Key: TEPHRA-270 > URL: https://issues.apache.org/jira/browse/TEPHRA-270 > Project: Tephra > Issue Type: Improvement > Components: core > Reporter: Poorna Chandra > Assignee: Poorna Chandra > Priority: Major > Fix For: 0.14.0-incubating > > > Transaction state cache, on startup, refreshes the transaction snapshot from > HDFS. The transaction state cache startup is called during HBase co-processor > startup. The refresh of the transaction snapshot can block, thus blocking the > co-processor startup. > The subsequent transaction snapshot refreshes happen in the background. Also > - there is no guarantee that a snapshot will be present during the refresh on > startup. Hence it is okay not to refresh the snapshot during snapshot, but do > it in background. -- This message was sent by Atlassian JIRA (v7.6.3#76005)