Details
-
Bug
-
Status: Closed
-
Trivial
-
Resolution: Fixed
-
None
-
None
Description
I saw this happening on tests related to autoscaling. The old directory clean up is triggered on core close in a separate thread. This can cause a race condition where the filesystem is closed before the cleanup starts running. Then a NPE is thrown and cleanup fails.
Fixing the NPE is simple but I think this is a real bug where old directories can be left around on HDFS. I don't know enough about HDFS to investigate further. Leaving it here for interested people to pitch in.
105029 ERROR (OldIndexDirectoryCleanupThreadForCore-control_collection_shard1_replica_n1) [n:127.0.0.1:58542_ c:control_collection s:shard1 r:core_node2 x:control_collection_shard1_replica_n1] o.a.s.c.HdfsDirectoryFactory Error checking for old index directories to clean-up. java.io.IOException: Filesystem closed at org.apache.hadoop.hdfs.DFSClient.checkOpen(DFSClient.java:808) at org.apache.hadoop.hdfs.DFSClient.listPaths(DFSClient.java:2083) at org.apache.hadoop.hdfs.DFSClient.listPaths(DFSClient.java:2069) at org.apache.hadoop.hdfs.DistributedFileSystem.listStatusInternal(DistributedFileSystem.java:791) at org.apache.hadoop.hdfs.DistributedFileSystem.access$700(DistributedFileSystem.java:106) at org.apache.hadoop.hdfs.DistributedFileSystem$18.doCall(DistributedFileSystem.java:853) at org.apache.hadoop.hdfs.DistributedFileSystem$18.doCall(DistributedFileSystem.java:849) at org.apache.hadoop.fs.FileSystemLinkResolver.resolve(FileSystemLinkResolver.java:81) at org.apache.hadoop.hdfs.DistributedFileSystem.listStatus(DistributedFileSystem.java:860) at org.apache.hadoop.fs.FileSystem.listStatus(FileSystem.java:1517) at org.apache.hadoop.fs.FileSystem.listStatus(FileSystem.java:1557) at org.apache.solr.core.HdfsDirectoryFactory.cleanupOldIndexDirectories(HdfsDirectoryFactory.java:540) at org.apache.solr.core.SolrCore.lambda$cleanupOldIndexDirectories$32(SolrCore.java:3019) at java.lang.Thread.run(Thread.java:745) 105030 ERROR (OldIndexDirectoryCleanupThreadForCore-control_collection_shard1_replica_n1) [n:127.0.0.1:58542_ c:control_collection s:shard1 r:core_node2 x:control_collection_shard1_replica_n1] o.a.s.c.SolrCore Failed to cleanup old index directories for core control_collection_shard1_replica_n1 java.lang.NullPointerException at org.apache.solr.core.HdfsDirectoryFactory.cleanupOldIndexDirectories(HdfsDirectoryFactory.java:558) at org.apache.solr.core.SolrCore.lambda$cleanupOldIndexDirectories$32(SolrCore.java:3019) at java.lang.Thread.run(Thread.java:745)
Attachments
Issue Links
- is related to
-
SOLR-13330 Improve HDFS tests
- Closed