HDDS-2295. Display log of freon on the standard output

[elek]

What changes were proposed in this pull request?

HDDS-2042 disabled the console logging for all of the ozone command line tools including freon.

But freon is different, it has a different error handling model. For freon we need all the log on the console.

1. To follow all the different errors

2. To get information about the used (random) prefix which can be reused during the validation phase.

I propose to restore the original behavior for Ozone.

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-2295

How was this patch tested?

Start any of the docker compose cluster and do a

ozone freon ockg -n10

You should see the logs (for example the notice about the used prefix)

[adoroszlai]

I'm not sure Freon is really that much different. It has the "progress bar" that should ideally not be messed up by the various log lines. Maybe the prefix should be printed to stdout? (I usually use one or more predefined prefixes, so I might be less sensitive to the importance of this. ;) )

Here's some sample output. I find it confusing/messy.

2019-10-14 17:44:20,616 INFO impl.MetricsConfig: Loaded properties from hadoop-metrics2.properties
2019-10-14 17:44:20,780 INFO impl.MetricsSystemImpl: Scheduled Metric snapshot period at 10 second(s).
2019-10-14 17:44:20,780 INFO impl.MetricsSystemImpl: ozone-freon metrics system started
2019-10-14 17:44:21,057 [main] INFO       - Executing test with prefix d0r64qhW41
2019-10-14 17:44:22,568 [main] INFO       - Creating Volume: vol1, with hadoop as owner.
2019-10-14 17:44:22,672 [main] INFO       - Creating Bucket: vol1/bucket1, with Versioning false and Storage Type set to DISK and Encryption set to false

 0.00% |?                                                                                                    |  0/100 Time: 0:00:002019-10-14 17:44:23,468 WARN impl.MetricsSystemImpl: ozone-freon metrics system already initialized!
 0.00% |?                                                                                                    |  0/100 Time: 0:00:01WARNING: An illegal reflective access operation has occurred
WARNING: Illegal reflective access by org.apache.ratis.thirdparty.com.google.protobuf.UnsafeUtil (file:/opt/hadoop/share/ozone/lib/ratis-thirdparty-misc-0.2.0.jar) to field java.nio.Buffer.address
WARNING: Please consider reporting this to the maintainers of org.apache.ratis.thirdparty.com.google.protobuf.UnsafeUtil
WARNING: Use --illegal-access=warn to enable warnings of further illegal reflective access operations
WARNING: All illegal access operations will be denied in a future release
 100.00% |?????????????????????????????????????????????????????????????????????????????????????????????????????|  100/100 Time: 0:00:08
10/14/19, 5:44:30 PM ===========================================================

-- Timers ----------------------------------------------------------------------
key-create
             count = 100
         mean rate = 12.32 calls/second
     1-minute rate = 8.00 calls/second
     5-minute rate = 8.00 calls/second
    15-minute rate = 8.00 calls/second
               min = 174.32 milliseconds
               max = 3420.12 milliseconds
              mean = 699.29 milliseconds
            stddev = 854.27 milliseconds
            median = 392.20 milliseconds
              75% <= 644.75 milliseconds
              95% <= 3167.49 milliseconds
              98% <= 3337.60 milliseconds
              99% <= 3388.52 milliseconds
            99.9% <= 3420.12 milliseconds


Total execution time (sec): 8
Failures: 0
Successful executions: 100

[elek]

I'm not sure Freon is really that much different. It has the "progress bar" that should ideally not be messed up by the various log lines.

It's developer tool, I think it's better to print out all the available development information. I am very happy to create a follow up jira and improve the progressbar. The current progressbar is very nice from cli but can't work very well in containerized environment.

Freon is like a long-live service not a cli app. Similar to scm or om, which prints out all the problems to the stdout.

[elek]

Here's some sample output. I find it confusing/messy.

Try the same with errors/problems ;-) If you really hate these information we can hide it in case of using a flag...

[adoroszlai]

The current progressbar is very nice from cli but can't work very well in containerized environment.

OK, I think I finally understand your problem.