Health monitor server status

[const-t]

Scope

Looks strange following condition in tfw_http_hm_control() if we take into account that tfw_apm_hm_srv_alive() must be called for alive server at least once to calculate valid crc32 in case when directive resp_crc32 is auto:

if (!tfw_srv_suspended(srv) ||
	    !tfw_apm_hm_srv_alive(resp->status, &resp->body, srv->apmref))

Discussion in PR

Need to find out is there a bug or not and develop functional test for this case.

[enuribekov-tempesta]

It looks code itself is correct, but number of issues was found around this code:
typos, bugs, unused function arguments etc.

[enuribekov-tempesta]

Created issue for bug found during testing HM.

#2128

[enuribekov-tempesta]

Sometime could't receive correct answer from monitored server.
The reason was tempesta-tech/tempesta-test#625
Reworked test for using nginx and got correct answer.

[enuribekov-tempesta]

tfw_http_hm_suspend() worked incorrectly, reworked.

[enuribekov-tempesta]

In a function tfw_apm_hm_srv_alive() hm->codes bitmap is empty, so always return error during comparation with any real status code.

[enuribekov-tempesta]

It looks like instead of saving server_failover_http configuration we got memory leak.
#2130

[krizhanovsky]

fixed in #2151