add stat_fork_rate in stat information

[jiafu1115]

@antirez Thanks for your review

[trevor211]

What do you think about it? @oranagra
Is it designed to report every stat variables to info stats?

[oranagra]

Truth be told, I never understood the purpose of this stat, the variable, its use in the LATENCY DOCTOR.
But i'm not sure it's a justification to remove it.
This PR is a bit shy on details..
@jiafu1115 please state why do you want to remove it.

[oranagra]

Ohh, i now notice this PR is trying to add a stat, not remove it.
the word "lose" in the title confused me.
still, @jiafu1115 please describe why you wanna add this stat?

looking closer, this variable seems to measure the time it takes the fork syscall to clone the page table (GB per second).

[jiafu1115]

Ohh, i now notice this PR is trying to add a stat, not remove it.
the word "lose" in the title confused me.
still, @jiafu1115 please describe why you wanna add this stat?

looking closer, this variable seems to measure the time it takes the fork syscall to clone the page table (GB per second).

@oranagra yes. the PR is to add the stat instead of to remove it. Redis has the variable and can monitor it so I suggest to add the stat similar with other ones with little effort and risk. After all, all variables except this one had/can be monitored except this one so I use "lose" word instead of "add" in this PR.
What's more, the stat_fork_rate is one of important monitored items when performance test tunning or production performance monitor. We often store so many keys into the redis and want to know all the performance items which may influence application when trouble shooing.

BTW: I update the title to avoid confusing us. thanks

[oranagra]

I don't see any value in that Stat, in order to add it, it needs to have a much better descriptive name, current name is very bad (which is ok for an internal variable that can be renamed anytime, but not for an API that needs to be steady).
Anyway, I don't see how this stat is useful, I think you're much better enable the latency monitor and you'll see for how long the fork freezes (in milliseconds rather than GB per second).

The fact that other stats variables are exposed is not a reason to add it.

[jiafu1115]

I don't see any value in that Stat, in order to add it, it needs to have a much better descriptive name, current name is very bad (which is ok for an internal variable that can be renamed anytime, but not for an API that needs to be steady).
Anyway, I don't see how this stat is useful, I think you're much better enable the latency monitor and you'll see for how long the fork freezes (in milliseconds rather than GB per second).

The fact that other stats variables are exposed is not a reason to add it.

got it. thank for your comments. "see for how long the fork freezes", what's the monitor item?