Misleading error message when locale mismatch occurs

[kikoreis]

Hello there. I've spent a bit of time today trying to figure out why mosh wouldn't work with a server farm I run and realized it's the same issue that many others have reported. I think the error output when mosh encounters this problem is somewhat bewildering because:

• The output is very long and refers to locales, a somewhat arcane topic
• The last line in the error output provides misleading information: /usr/bin/mosh: Did not find mosh server startup message. (Have you installed mosh on your server?)

I'll paste the raw output into my next comment to stay on point; I think the output should, ideally, be a one-liner, something like:

/usr/bin/mosh: server missing files for client locale pt_BR.UTF-8; to fix generate server locales or modify client locale

This would be in line with the idea of providing a general error message in the last line, such as what mosh displays when the server is indeed not installed.

[kikoreis]

The raw output is 31 lines, so if you're skimming and only focus on the last line as I was you might think the top part is a warning and something else is wrong:

The locale requested by LC_CTYPE=pt_BR.UTF-8 isn't available here.
Running `locale-gen pt_BR.UTF-8' may be necessary.

The locale requested by LC_CTYPE=pt_BR.UTF-8 isn't available here.
Running `locale-gen pt_BR.UTF-8' may be necessary.

mosh-server needs a UTF-8 native locale to run.

Unfortunately, the local environment (LC_CTYPE=pt_BR.UTF-8) specifies
the character set "US-ASCII",

The client-supplied environment (LC_CTYPE=pt_BR.UTF-8) specifies
the character set "US-ASCII".

locale: Cannot set LC_CTYPE to default locale: No such file or directory
locale: Cannot set LC_ALL to default locale: No such file or directory
LANG=en_US.UTF-8
LANGUAGE=
LC_CTYPE=pt_BR.UTF-8
LC_NUMERIC=pt_BR.UTF-8
LC_TIME=pt_BR.UTF-8
LC_COLLATE="en_US.UTF-8"
LC_MONETARY=pt_BR.UTF-8
LC_MESSAGES="en_US.UTF-8"
LC_PAPER=pt_BR.UTF-8
LC_NAME=pt_BR.UTF-8
LC_ADDRESS=pt_BR.UTF-8
LC_TELEPHONE=pt_BR.UTF-8
LC_MEASUREMENT=pt_BR.UTF-8
LC_IDENTIFICATION=pt_BR.UTF-8
LC_ALL=
Connection to morphys closed.
/usr/bin/mosh: Did not find mosh server startup message. (Have you installed mosh on your server?)

[kikoreis]

I think this is worth fixing as that this seems to be a very common problem, as there are many issues reported here, including #1112, #1137, #1136, #793, #98, #678, #1096.

If users are concerned with regenerating server locales (which I believe is a valid concern), the trivial one-liner that works with most mismatches is to invoke mosh as LC_ALL=C.UTF-8 mosh but that's not obvious; the problem is made somewhat worse due to the fact that the default sshd config on Ubuntu specifies SendEnv LANG LC_*, which many users get confused about.

I'm happy to add text for an FAQ to complement the shorter error message.

[achernya]

If users are concerned with regenerating server locales (which I believe is a valid concern), the trivial one-liner that works with most mismatches is to invoke mosh as LC_ALL=C.UTF-8 mosh but that's not obvious

This is not guaranteed to work either. Not all systems have a C.UTF-8 locale. Unfortunately this is not something that mosh can fix generally. Mosh requires a UTF-8 locale, and end-users may not have privileges on those hosts to install a locale, nor is locale generation a unified command on all platforms where mosh runs.

Personally, I think the error message /usr/bin/mosh: Did not find mosh server startup message. is pretty reasonable; the (Have you installed mosh on your server?) bit was added as that's the 99% case for why mosh server couldn't be started. The locale issues are (relatively) rare, but the next most common issue.

If you (or anyone else) would like to offer up a pull request to mosh.pl to detect locale errors and print a better parenthetical, I'm happy to review it.

[kikoreis]

Yeah, I realize LC_ALL=C.UTF-8 is not a panacea, but it does work for the most common user profile that hits this bug, a client configured with non-US locale attempting to connect to Ubuntu server with an unmodified locale configuration. On a typical, fresh 24.04 LTS VM, you get:

ubuntu@morphys:~$ locale -a
C
C.utf8
POSIX
en_US.utf8

which is why so many people hit this. Let me look at the source to see if I can figure out how to produce a patch.

[achernya]

According to #1137 (comment) macOS does not support C.UTF-8. locale -a does print C.UTF-8 on my mac, but /usr/share/locale/C.UTF-8 only contains LC_CTYPE (unlike en_US.UTF-8, which has a bunch), so it might not be a fully usable locale? Empirical testing does seem to show it works.

Nonetheless, we cannot (from a mosh perspective) assume that a user will have a C.UTF-8 locale, nor even an en_US.UTF-8 locale. But an improved error message is certainly a possibility.