Cassandra fails with the "java.lang.OutOfMemoryError: Direct buffer memory" error
With some large databases, Cassandra may fail with the following error:
java.lang.OutOfMemoryError: Direct buffer memory
at java.nio.Bits.reserveMemory(Bits.java:694)
at java.nio.DirectByteBuffer.<init>(DirectByteBuffer.java:123)
at java.nio.ByteBuffer.allocateDirect(ByteBuffer.java:311)
at org.apache.cassandra.utils.memory.BufferPool.allocate(BufferPool.java:110)
at org.apache.cassandra.utils.memory.BufferPool.access$1000(BufferPool.java:46)
at org.apache.cassandra.utils.memory.BufferPool$LocalPool.allocate(BufferPool.java:407)
at org.apache.cassandra.utils.memory.BufferPool$LocalPool.access$000(BufferPool.java:334)
at org.apache.cassandra.utils.memory.BufferPool.takeFromPool(BufferPool.java:122)
at org.apache.cassandra.utils.memory.BufferPool.get(BufferPool.java:94)
at org.apache.cassandra.cache.ChunkCache.load(ChunkCache.java:155)
at org.apache.cassandra.cache.ChunkCache.load(ChunkCache.java:39)
To solve the problem do one of the following
Increase the memory (Xmx VM option) for Cassandra until the migration is successful.
Set "file_cache_size_in_mb: 0" In the cassandra.yaml file to disable the chunk cache completely.
"All commit element IDs" index is corrupted
If the "all commit element IDs" index is corrupted, migration may fail and print and error of the following format into the Migrator's log file:
It has been detected that "all commit element IDs" index is corrupted at revision <revision> of resource <resource.id>
To solve the problem
Wait until all selected projects finish migrating.
Stop the Migrator (backend) process.
At the end of the migrator's application.conf file, add the following line: esi.migration.source-revision-all-object-ids-getter.prefer-bfs=true
Start the Migrator and retry the migration of the failed project from the Migrator GUI.
With this setting in place, it is expected that the migration progress will be much slower so it is recommended to finish the affected project with this setting and then revert back.
The "java.net.SocketException: Too many open files" error appears during migration
This error usually means that the Cassandra or Teamwork Cloud process is running into system-imposed limits on a number of open files.
To solve the problem
Set the following limits in /etc/security/limits.conf. localUser hard nofile 50000 localUser hard nproc 50000 localUser soft nofile 40000 localUser soft nproc 40000 cassandra hard nofile 50000 cassandra hard nproc 50000 cassandra soft nofile 40000 cassandra soft nproc 40000 root hard nofile 50000 root hard nproc 50000 root soft nofile 40000 root soft nproc 40000
Restart the processes.
Limits are configurable to separate users (like the Cassandra user that runs Cassandra or the Teamwork Cloud user that runs the Teamwork Cloud service). For more information, see https://linux.die.net/man/5/limits.conf.
The "com.datastax.driver.core.exceptions.AuthenticationException" exception is shown in the log file
If the The following exception is shown in the log file:
Error while trying to activate [com.nomagic.esi.server.core.a.d.c, main] com.datastax.driver.core.exceptions.AuthenticationException: Authentication error on host FQDN/IP: Provided username cassandra and/or password are incorrect
To solve the problem
Add the following configuration block to application.conf of the migrator artefact: