producer/transport connection redundant logs

[baallan]

As in v3, v4 needs connection-related log messages related to reasonably-expected failures to be issued once per incident instead of repeated at the attempt interval. Logs shouldn't explode because a sampled node is down.

[tom95858]

Hi @baallan, please propose a design to be discussed.

[baallan]

Barring a refactoring of logging in v4, I would add a log_status int in struct ldmsd_prdcr and a set of bit flags (e.g. PRDCR_LOG_REJECTED, PRDCR_LOG_ERROR) handled such that if the bit is set, the message is suppressed (having already been issued). A successful connection should clear the flag(s).