you are now in the presence of someone who has done something widely considered impossible in the IT field: speaking to a responsible party, at Google, for Gmail, getting them to acknowledge an issue, and work it as a P1 ticket.
so, we were rolling out a new mail transit agent and had the concurrency settings wrong (relying only on retry delay, not categorizing by recipient domain), but not in a way that was causing us a problem... until it did. gmail started applying rate limits, started ratcheting them tighter, across the entire goddamn domain instead of the very clearly demarcated-in-the-rDNS deployment boundaries, in some kind of closed-loop-nobody-paying-attention situation. also their postmaster-tools dashboards lag by 48h, making it fundamentally impossible to gauge the effects of our changes on a reasonable timeline. then one of the config options we apply on monday causes a show-stopper concurrency bug in the mail server package and we get paged at 0am on tuesday because the queue is wedged. executives get involved, the people that manage our relationship with Google promise that they'll schedule a meeting. and then we get paged last night at midnight again, and had to say 'we have done everything we can to control this situation with the configs and data we have, paging me won't make the mail flow any faster, or make the scheduled call happen any sooner, or make them respect our deployment boundaries'
so we get on a call and repeat basically all of this to the google rep and add "this behavior on your end does not reflect an understanding of the realities of running a large SMTP service on the internet"
they admitted fault and cleared the rate limits
