Support log rotate

[spikeekips]

Github Issue

Resolves #873

Background

In debug mode, SEBAK produces huge log messages including HTTP requests and responses log. Under docker, we can use max-size setting in log driver, but a node can keep the entire log messages, it's the matter of choice.

Solution

There are 2 logs, node log, and HTTP log. Each log will have rotate options,

• --log-rotate-max-size, --http-log-rotate-max-size: if log file is over, rotate log
• --log-rotate-max-age, --http-log-rotate-max-age: if log file is over days, rotate log
• --log-rotate-max-count, --http-log-rotate-max-count: if the number of rotated logs over, remove olds
• --log-rotate-uncompress, --http-log-rotate-uncompress: compress or not

[codecov-io]

Codecov Report

Merging #886 into master will decrease coverage by 0.39%.
The diff coverage is 28.4%.

@@            Coverage Diff            @@
##           master     #886     +/-   ##
=========================================
- Coverage   60.42%   60.02%   -0.4%     
=========================================
  Files         153      153             
  Lines       10774    10833     +59     
=========================================
- Hits         6510     6503      -7     
- Misses       3521     3585     +64     
- Partials      743      745      +2

Flag	Coverage Δ
#integration_tests_long_term	44.24% <28.4%> (-0.3%)	⬇️
#integration_tests_node	40.25% <28.4%> (-0.19%)	⬇️
#unittests	48.43% <26.13%> (-0.19%)	⬇️

Impacted Files	Coverage Δ
lib/common/log.go	3.03% <0%> (-0.55%)	⬇️
cmd/sebak/cmd/run.go	54.56% <32.05%> (-2.97%)	⬇️
lib/sync/fetcher.go	57.5% <0%> (-7.5%)	⬇️
lib/sync/try.go	78.57% <0%> (-7.15%)	⬇️
lib/metrics/sync.go	88.57% <0%> (-5.72%)	⬇️
lib/network/validator_connection_manager.go	84.18% <0%> (-0.8%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e1ae687...cbe6eba. Read the comment docs.

[Geod24]

I thought you wanted to use logrorate directly ? At least that's what we discussed in the issue

[spikeekips]

@Geod24 Hmm, logrotate utility is one of our options. It can be a simple solution to support a rotating log, but under docker, environment to make logrotate to be working needs some additional efforts. I think the rotating log should be the native feature for considering the various deploying environments.

[Geod24]

@spikeekips : What kind of additional effort ? It should be a few (<10) LoC in sebak and a configuration file.

[spikeekips]

@Geod24 :) For instance, like these
https://www.google.co.kr/search?q=do+ker+logrotate&ie=UTF-8&oe=UTF-8&hl=en-kr&client=safar . Basicslly docker runs one entrypoint, so the daemon like logrotate should be deployed by task manager in docker.