meta.math.stackexchange.com is sometimes reasonable, but far too often toxic.
Bullying by cliques on meta
The "meta" sites are designed in a way that enables bullying and boorishness. There is a clique of users with no official positions who have decided that some topics are forbidden, and they close all comments on those topics on bogus pretexts, and are extraordinary harsh and angry for no apparent reason in the way they address anyone who does not bow down to them in a servile fashion.
When this is reported to the moderators, the moderators make no comment and just ignore it. When it was reported dozens of times over several years they made no comment and they ignored the reports.
I don't think this will get fixed until two things happen: (1) No voting on postings to "meta" can be done, and (2) postings to "meta" will not be closed without discussion, and such discussion will be an actual discussion rather than phrased like instructions to a plantation slave.
Frequent drive-by disrespect involving insinuations that can be easily and almost instantly seen to be false
Besides bullying by cliques, another sort of commonplace rude behavior on "meta" sites is this: If you say something is wrong with anything about the way things are conventionally done here, someone instantly responds by saying maybe you're a newbie who doesn't know this or that (specified—in every instance there is specificity) or who made this or that particular mistake (always specified) and they ignore the fact that the poster has a reputation above 200 000 (two-hundred-thousand) and the fact that the person has not committed any of the specified mistakes and is not ignorant of the specified things.
Then typically within one minute, the contemptuous comment has a dozen or more upvotes.
Abuse built in to the software, now fixed
On math.stackexchange.com one extraordinarily boorish thing was built in to the software and has recently been fixed many years after I proposed the particular software fix.
That is simply that a posting could not be closed for insufficient context without an official notice to the poster saying the posting is not about mathematics, when obviously it is. This bug was finally fixed after many years.
Another form of disrespect built in to the software
Another instance of disrespect built in to the software is in the close queue, where people vote on whether to close questions. Users are not allowed to look over the questions whose closing is proposed and judiciously choose which ones they are able to contribute to.
Rather they are told: Here's the next question you are to look at. You don't get to see other questions whose closing is proposed until you've either voted and comment on this one or clicked on "skip", and if you click on "skip", you will not be allowed to choose which question to work on next, but instead you will be told.
Only people willing to submit to this treatment can habitually work on the close queue, and I wonder if that is why those who habitually work on the close queue feel entitled, as in fact they do, to be disrespectful to others.
Actual on-topic postings are nonetheless usually polite
Material that's actually on the topic to which the site is devoted—the questions and answers—is almost always polite and respectful and businesslike and sometimes friendly.