Closed egli closed 9 years ago
Seems like other people had this problem before: https://bz.apache.org/bugzilla/show_bug.cgi?id=45786#c3
I still get
2015-04-30 12:14:43,371 ERROR [mdr2.util] (Thread-1307 (HornetQ-client-global-threads-410738640)) #<DirectoryNotEmptyException java.nio.file.DirectoryNotEmptyException: /var/lib/mdr2/recorded/20127>
in the log
For now I increase the retry sleep. Let's see if that helps. Maybe in the end we'll just resort to
(sh "rm" "-rf" directory)
Now it is slightly better but still totally unacceptable:
2015-04-30 17:00:55,418 ERROR [mdr2.util] (Thread-35 (HornetQ-client-global-threads-1292880237)) #<FileSystemException java.nio.file.FileSystemException: /var/lib/mdr2/recorded/20131/.nfs000000008c17288c000008fd: Device or resource busy>
Haven't seen this problem in production anymore
Under some circumstances we get the following entry in the log:
This seems to be a problem in conjunction with nfs as there are some
.nfsxxxxxxx
files in the directory. Maybe we need to put a delay in the tree walker that does the delete? Or we could have a reaper that regularly cleans directories of archived productions.