Resolve memory leak in redisson instrumentation #10288
Conversation
AlchemyDing
changed the title
| if (commandsData.getCommands().size() == 1) { | ||
| return commandsData.getCommands().get(0).getCommand().getName(); | ||
| } else { | ||
| return "BATCH EXECUTE"; |
There was a problem hiding this comment.
Does the spec allow this? From what I understand from https://github.com/open-telemetry/semantic-conventions/blob/main/docs/database/database-spans.md#call-level-attributes operation should not be set if there is more than one operation.
There was a problem hiding this comment.
db.operation.name
I noticed that there is a description of the db.operation.name for batch statements in semantic-conventions, but I'm not sure if this is the one adjusted after this pull request
| default: | ||
| return String.join(";", sanitizedStatements); | ||
| } | ||
| return sanitizeStatement(); |
There was a problem hiding this comment.
instead of omitting the statement when multiple statements are executed you could have changed it to only capture some small number X statements or limit the total length of the statement (there is a size limit to attributes anyway). Did you consider solving it this way?
|
@AlchemyDing Just curious is this a guess that this change will solve the OOM or are you able to repro it? |
# Conflicts: # instrumentation/redisson/redisson-common/javaagent/src/main/java/io/opentelemetry/javaagent/instrumentation/redisson/RedissonRequest.java # instrumentation/redisson/redisson-common/testing/src/main/java/io/opentelemetry/javaagent/instrumentation/redisson/AbstractRedissonAsyncClientTest.java # instrumentation/redisson/redisson-common/testing/src/main/java/io/opentelemetry/javaagent/instrumentation/redisson/AbstractRedissonClientTest.java
2 participants
resolve issue
When a large number of commands are sent to the Redis server at once using the
RedisConnection.send(CommandsData data)method, a significant number ofnormalizeSingleCommandoperations are executed. This can result in a long statement that takes up a large amount of memory or causes an Out-of-Memory (OOM) error. When users use batch execute, they usually do not pay attention to the specific content of each statement.