Run OTEL collector without configuration file #2148
Conversation
Signed-off-by: Pavol Loffay <ploffay@redhat.com>
Signed-off-by: Pavol Loffay <ploffay@redhat.com>
|
@annanay25 would you like to take this one? |
Codecov Report
@@ Coverage Diff @@
## master #2148 +/- ##
==========================================
+ Coverage 96.12% 96.14% +0.02%
==========================================
Files 218 219 +1
Lines 10577 10585 +8
==========================================
+ Hits 10167 10177 +10
+ Misses 354 352 -2
Partials 56 56
Continue to review full report at Codecov.
|
Signed-off-by: Pavol Loffay <ploffay@redhat.com>
Signed-off-by: Pavol Loffay <ploffay@redhat.com>
| } else if storage == "kafka" { | ||
| kaf := factories.Exporters[kafka.TypeStr].CreateDefaultConfig() | ||
| exporters[kafka.TypeStr] = kaf | ||
| } |
There was a problem hiding this comment.
Can we use a switch case here and error out on a bad storage type here, instead of passing it down to main.go and handling it in storageFlags()
There was a problem hiding this comment.
+1 on the switch, the storageFlags from main also parses the variable but it has a different functionality
| func createProcessors(factories config.Factories) configmodels.Processors { | ||
| batch := factories.Processors["batch"].CreateDefaultConfig().(*batchprocessor.Config) | ||
| return map[string]configmodels.Processor{ | ||
| "batch": batch, |
There was a problem hiding this comment.
How did we decide this list of default processors? I'd like to think queued_retry is also part of this list as it reduces data loss. Which brings about a more important question of - what are the use cases of running the otel-collector without a configuration file? In what scenario would we want to do that?
There was a problem hiding this comment.
Good point - the other reason for using queued_retry is that it will have some of the metrics we are interested in using.
There was a problem hiding this comment.
I have used the batch as a starting point since there has to be at least one processor. In the production-ready settings we might run batch, retry and memory limiter. Let's handle this in a separate task.
Which brings about a more important question of - what are the use cases of running the otel-collector without a configuration file? In what scenario would we want to do that?
The highest priority we had was to be a drop-in replacement for the current collector. The OTEL collector fails without the config file which is not compatible behavior with our collector.
There was a problem hiding this comment.
There was a problem hiding this comment.
there has to be at least one processor
IIRC, this is no longer true, but yes I agree, let's start with the batch processor and merge this soon.
| storageType := os.Getenv(storage.SpanStorageTypeEnvVar) | ||
| if storageType == "" { | ||
| storageType = "cassandra" | ||
| } | ||
|
|
There was a problem hiding this comment.
Seems like it belongs better in the config file.
There was a problem hiding this comment.
I would like to keep the parsing of flags and getting the settings from the environment in the main - in one place.
Signed-off-by: Pavol Loffay <ploffay@redhat.com>
SPAN_STORAGE_TYPEenv variable and by default Cassandra implementation is used.Signed-off-by: Pavol Loffay ploffay@redhat.com