Running yb-admin and other tools produces a few lines of extra output

[mbautin]

Jira Link: DB-1737
yb-admin outputs the following regardless of the command:

I0328 17:21:27.967152 15789 flags.cc:37] Auto setting FLAGS_yb_num_shards_per_tserver to 8
I0328 17:21:27.968122 15789 mem_tracker.cc:158] MemTracker: hard memory limit is 12.483986 GB
I0328 17:21:27.968154 15789 mem_tracker.cc:160] MemTracker: soft memory limit is 10.611388 GB

[bmatican]

Also happens in PG logs on new connections.

I0705 13:17:49.294610 30967168 mem_tracker.cc:299] MemTracker: hard memory limit is 10.400000 GB
I0705 13:17:49.294703 30967168 mem_tracker.cc:301] MemTracker: soft memory limit is 8.840000 GB
I0705 13:17:49.295032 30967168 thread_pool.cc:170] Starting thread pool { name: pggate_ybclient queue_limit: 10000 max_workers: 1024 }

Since this comes from the root tracker, we could probably just default to not logging anything and just allow a one time logging in master & tserver.

[joe-maley]

1. I realize this ticket is 4+ years old. The yb-admin tool no longer outputs the extra output by default. I am only able to reproduce the memory limit reporting by using the -minloglevel 1 flag.

2. The memory limit reporting happens in the root tracker. As Bogdan suggested, we can avoid reporting the memory limits in the yb-admin and PG logs by only reporting these stats when the master and tserver classes are constructed.

3. The Starting thread pool … log is just a standard log statement that is generated whenever a ThreadPool is constructed. We could do something like introducing a flag into the ThreadPoolOptions to control whether or not we print this log statement, but I suspect that this issue isn’t really bloating the code with such a rigid fix. What do you think @bmatican?

[joe-maley]

Fixed by 0d6e495