3 ms·

> #1: Disastrous XID Wraparound > Pretty much any non-trivial PostgreSQL install that isn’t staffed with a top expert will run into it eventually. I agree tha

> #1: Disastrous XID Wraparound > Pretty much any non-trivial PostgreSQL install that isn’t staffed with a top expert will run into it eventually. I agree that this landmine is particularly nasty - and I think it needs to be fixed upstream somehow. But I do think it is fairly well known at this point. Or at least, people outside of "top expert[s]" have heard of it and are at least aware of the problem by now.

dfox 6y ago
In normal use XID wraparound is not particularly problematic. For it to be an issue you have to either have application that for some reason assumes that XIDs monotonically increase (for example in its implementation of optimistic locking) or you have significantly larger issue with totally unmanaged MVCC bloat (caused by either not running vacuum at all or by really having ridiculous amount of ridiculously long-running transactions active at once). But then there is interesting related issue in some client libraries: the XID is 32bit unsigned value and some libraries which transparently implement optimistic locking (eg. ODBC) interpret it as 32b signed value. I somewhat suspect that most people who had "significant production outage caused by XID wraparound" were in fact bitten by this or something similar.