setlogginglevel sets the level log threshold for a given component or class during runtime. If this command is called with no parameters, the log level is reset to the initial configuration.
Note
using trace or debug logging levels create very large log files where the reader may not find what they are looking for. It is best to use these levels for a very short period of time or with the help of Scylla Support.
nodetool setlogginglevel <class> <level>
Where class is the specific class you want a log for, and level is one of the below levels.
Example
This example sets the logging level of the paging class to trace:
nodetool setlogginglevel paging trace
You can set the log to any of the following log threshold levels presented here in order of severity and verbosity:
trace
debug
info - default setting
warn
error
To display the log classes (output changes with each version so your display may be different).
scylla --help-loggers
lsa-api
trace_state
tracing
password_authenticator
init
stream_session
lister
load_broadcaster
paging
storage_service
org.apache.cassandra.db.marshal.TimestampType
scollectd
event_notifier
thrift
org.apache.cassandra.db.marshal.DateType
collection_type_impl
repair
compaction
lsa
failure_injector
BatchStatement
seastar_memory
migration_task
legacy_schema_migrator
sstable
dns_resolver
lsa-timing
database
cql_server
storage_proxy
cache
DateTieredCompactionStrategy
schema_tables
rpc
compaction_manager
seastar
migration_manager
trace_keyspace_helper
query_processor
gossip
failure_detector
schema_registry
replication_strategy
LeveledManifest
mutation_data
consistency
commitlog
default_authorizer
auth
messaging_service
config
commitlog_replayer
heat_load_balance
batchlog_manager
token_metadata
view
bloom_filter
unimplemented
boot_strapper
query_result
system_keyspace
ange_streamer