server: add Uptime status counter

[morgo]

What problem does this PR solve?

Fixes #8842

What is changed and how it works?

Uses variable infrastructure to push Uptime statistic. So you can now do this, as in MySQL:

tidb> show global status like 'Uptime';
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| Uptime        | 10    |
+---------------+-------+
1 row in set (0.00 sec)

Check List

Tests

• Manual test

I would like to add a test, but I wanted to check first - since the logical way to do it is to run show global status, sleep a second, and then run it again. But I don't know if:

1. Sleeping for 1 second is considered acceptable since it slows down the tests.
2. It handles the race condition successfully that sleeping is never exactly 1 second.

As an alternative, I could assert that the value is greater than zero? Feedback welcome.

Code changes

• Has exported function/method change
• Has exported variable/fields change
• Has interface methods change

Side effects

• Increased code complexity

Related changes

• Need to update the documentation (supported statuses, variables)
• Need to be included in the release note

---

This change is 

[codecov-io]

Codecov Report

Merging #8888 into master will increase coverage by <.01%.
The diff coverage is 25%.

@@            Coverage Diff             @@
##           master    #8888      +/-   ##
==========================================
+ Coverage   67.25%   67.25%   +<.01%     
==========================================
  Files         372      372              
  Lines       76301    76203      -98     
==========================================
- Hits        51316    51253      -63     
+ Misses      20453    20418      -35     
  Partials     4532     4532

Impacted Files	Coverage Δ
server/stat.go	0% <0%> (ø)	⬆️
server/server.go	53.15% <100%> (+0.02%)	⬆️
ddl/delete_range.go	74.28% <0%> (-4.58%)	⬇️
executor/join.go	78.44% <0%> (-1.04%)	⬇️
executor/distsql.go	72.53% <0%> (-0.47%)	⬇️
util/systimemon/systime_mon.go	100% <0%> (+20%)	⬆️

---

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 41838ce...4b47b61. Read the comment docs.

[wuudjac]

@morgo How about testing like this:

server, err := NewServer(...)
// Initialization stuff...

server.startTime = server.startTime.Add(-10 * time.Second)

result, err := server.Stats(...)
// Assert `result` got is not less than 10.

Since you are calculating duration, so you can change the lower bound instead of the upper bound.

For case 1, it will not slow down the tests.
For case 2, you can assert result is not less than any value you set, but cannot assert the upper limit in race condition.

If you want to assert upper limit, or want exact equal assertion, maybe you need to implement a mock time producer that offers time.Now()

[zz-jason]

Suggested change

	// Copyright 2018 PingCAP, Inc.
	// Copyright 2019 PingCAP, Inc.

[zz-jason]

@morgo friendly ping, any update?

[sre-bot]

Hi contributor, thanks for your PR.

This patch needs to be approved by someone of admins. They should reply with "/ok-to-test" to accept this PR for running test automatically.

[zz-jason]

@jackysp Since we have monitored the uptime on Grafana, maybe this PR is not that necessary?

[jackysp]

This PR still has some compatibility improvements. But not so important. Maybe we can close it first? @zz-jason

[zz-jason]

I'm going to close this PR due to outdated, fell free to reopen it if you are going to process it again in the future.