[jira] [Commented] (HBASE-8445) regionserver can't load an updated coprocessor jar with the same jar path
[ https://issues.apache.org/jira/browse/HBASE-8445?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13644099#comment-13644099 ] Jimmy Xiang commented on HBASE-8445: In that case, since you need to re-enable the table anyway, it's better for you to change the coprocessor jar name/path, so a difference class loader will be used. regionserver can't load an updated coprocessor jar with the same jar path - Key: HBASE-8445 URL: https://issues.apache.org/jira/browse/HBASE-8445 Project: HBase Issue Type: Bug Affects Versions: 0.94.5 Reporter: Wang Qiang Attachments: patch_20130426_01.txt when I update a coprocessor jar, then I disable and enable the table with the coprocessor, but the new features in the updated coprocessor jar doesn't make any sense. Follow into the class 'org.apache.hadoop.hbase.coprocessor.CoprocessorHost', I found that there's a coprocessor class loader cache , of which the key is the coprocessor jar path(although the key is a weak reference), so when I disable/enable the table, it got a cached coprocessor class loader from the cache with the jar path, and it didn't try to reload the coprocessor jar from the hdfs. Here I give a patch, in which I add an extra info which is 'FileCheckSum' with the coprocessor class loader cache, if the checksum is changed, try to reload the jar from the hdfs path -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-8445) regionserver can't load an updated coprocessor jar with the same jar path
[ https://issues.apache.org/jira/browse/HBASE-8445?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13643019#comment-13643019 ] Jimmy Xiang commented on HBASE-8445: That's right: we don't do reloading. One work-around is to do a full cluster rolling-restart in such a case. regionserver can't load an updated coprocessor jar with the same jar path - Key: HBASE-8445 URL: https://issues.apache.org/jira/browse/HBASE-8445 Project: HBase Issue Type: Bug Affects Versions: 0.94.5 Reporter: Wang Qiang Attachments: patch_20130426_01.txt when I update a coprocessor jar, then I disable and enable the table with the coprocessor, but the new features in the updated coprocessor jar doesn't make any sense. Follow into the class 'org.apache.hadoop.hbase.coprocessor.CoprocessorHost', I found that there's a coprocessor class loader cache , of which the key is the coprocessor jar path(although the key is a weak reference), so when I disable/enable the table, it got a cached coprocessor class loader from the cache with the jar path, and it didn't try to reload the coprocessor jar from the hdfs. Here I give a patch, in which I add an extra info which is 'FileCheckSum' with the coprocessor class loader cache, if the checksum is changed, try to reload the jar from the hdfs path -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-8445) regionserver can't load an updated coprocessor jar with the same jar path
[ https://issues.apache.org/jira/browse/HBASE-8445?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13643499#comment-13643499 ] Wang Qiang commented on HBASE-8445: --- in the version before 0.94.2, when I updated a coprocessor jar, it can make sense after I restart the table. 'do a full cluster rolling-restart in such a case' is not acceptable in the case that I only update a coprocessor jar for one table regionserver can't load an updated coprocessor jar with the same jar path - Key: HBASE-8445 URL: https://issues.apache.org/jira/browse/HBASE-8445 Project: HBase Issue Type: Bug Affects Versions: 0.94.5 Reporter: Wang Qiang Attachments: patch_20130426_01.txt when I update a coprocessor jar, then I disable and enable the table with the coprocessor, but the new features in the updated coprocessor jar doesn't make any sense. Follow into the class 'org.apache.hadoop.hbase.coprocessor.CoprocessorHost', I found that there's a coprocessor class loader cache , of which the key is the coprocessor jar path(although the key is a weak reference), so when I disable/enable the table, it got a cached coprocessor class loader from the cache with the jar path, and it didn't try to reload the coprocessor jar from the hdfs. Here I give a patch, in which I add an extra info which is 'FileCheckSum' with the coprocessor class loader cache, if the checksum is changed, try to reload the jar from the hdfs path -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira