We don't know why a few Windows users have seen this fail, but the taciturnity of the error message certainly isn't helping debug it. Let's at least find out which LC category isn't working.
When we invalidate the relcache entry for a system catalog or index, we must also delete the relcache "init file" if the init file contains a copy of that rel's entry. The old way of doing this relied on a specially maintained list of the OIDs of relations present in the init file: we made the list either when reading the file in, or when writing the file out. The problem is that when writing the file out, we included only rels present in our local relcache, which might have already suffered some deletions due to relcache inval events. In such cases we correctly decided not to overwrite the real init file with incomplete data --- but we still used the incomplete initFileRelationIds list for the rest of the current session. This could result in wrong decisions about whether the session's own actions require deletion of the init file, potentially allowing an init file created by some other concurrent session to be left around even though it's been made stale. Since we don't support changing the schema of a system catalog at runtime, the only likely scenario in which this would cause a problem in the field involves a "vacuum full" on a catalog concurrently with other activity, and even then it's far from easy to provoke. Remarkably, this has been broken since 2002 (in commit 786340441706ac1957a031f11ad1c2e5b6e18314), but we had never seen a reproducible test case until recently. If it did happen in the field, the symptoms would probably involve unexpected "cache lookup failed" errors to begin with, then "could not open file" failures after the next checkpoint, as all accesses to the affected catalog stopped working. Recovery would require manually removing the stale "pg_internal.init" file. To fix, get rid of the initFileRelationIds list, and instead consult syscache.c's list of relations used in catalog caches to decide whether a relation is included in the init file. This should be a tad more efficient anyway, since we're replacing linear search of a list with ~100 entries with a binary search. It's a bit ugly that the init file contents are now so directly tied to the catalog caches, but in practice that won't make much difference. Back-patch to all supported branches.
We should set MyProc->databaseId after acquiring the per-database lock, not beforehand. The old way risked deadlock against processes trying to copy or delete the target database, since they would first acquire the lock and then wait for processes with matching databaseId to exit; that left a window wherein an incoming process could set its databaseId and then block on the lock, while the other process had the lock and waited in vain for the incoming process to exit. CountOtherDBBackends() would time out and fail after 5 seconds, so this just resulted in an unexpected failure not a permanent lockup, but it's still annoying when it happens. A real-world example of a use-case is that short-duration connections to a template database should not cause CREATE DATABASE to fail. Doing it in the other order should be fine since the contract has always been that processes searching the ProcArray for a database ID must hold the relevant per-database lock while searching. Thus, this actually removes the former race condition that required an assumption that storing to MyProc->databaseId is atomic. It's been like this for a long time, so back-patch to all active branches.
Also sneak entries for commits 97ff2a564 et al into the sections for the previous releases in the relevant branches. Those fixes did go out in the previous releases, but missed getting documented.
The argument that this is a sufficiently-expected case to be silently ignored seems pretty thin. Andres had brought it up back when we were still considering that most fsync failures should be hard errors, and it probably would be legit not to fail hard for ETXTBSY --- but the same is true for EROFS and other cases, which is why we gave up on hard failures. ETXTBSY is surely not a normal case, so logging the failure seems fine from here.
Commit 2ce439f3379aed857517c8ce207485655000fc8e introduced a rather serious regression, namely that if its scan of the data directory came across any un-fsync-able files, it would fail and thereby prevent database startup. Worse yet, symlinks to such files also caused the problem, which meant that crash restart was guaranteed to fail on certain common installations such as older Debian. After discussion, we agreed that (1) failure to start is worse than any consequence of not fsync'ing is likely to be, therefore treat all errors in this code as nonfatal; (2) we should not chase symlinks other than those that are expected to exist, namely pg_xlog/ and tablespace links under pg_tblspc/. The latter restriction avoids possibly fsync'ing a much larger part of the filesystem than intended, if the user has left random symlinks hanging about in the data directory. This commit takes care of that and also does some code beautification, mainly moving the relevant code into fd.c, which seems a much better place for it than xlog.c, and making sure that the conditional compilation for the pre_sync_fname pass has something to do with whether pg_flush_data works. I also relocated the call site in xlog.c down a few lines; it seems a bit silly to be doing this before ValidateXLOGDirectoryStructure(). The similar logic in initdb.c ought to be made to match this, but that change is noncritical and will be dealt with separately. Back-patch to all active branches, like the prior commit. Abhijit Menon-Sen and Tom Lane
According to recent tests, this case now works fine, so there's no reason to reject it anymore. (Even if there are still some OpenBSD platforms in the wild where it doesn't work, removing the check won't break any case that worked before.) We can actually remove the entire test that discovers whether libpython is threaded, since without the OpenBSD case there's no need to know that at all. Per report from Davin Potts. Back-patch to all active branches.