Skip to content

fix(refresh_token): truncate token_out before overwriting it

Alexander Rogovskiy had to update the refresh token of our etf OAuth2 client and noticed the following issue:

06/05/24 08:33:58 Looking for token in file /opt/omd/sites/etf/etc/nagios/globus/lhcb-ce.token
06/05/24 08:33:58 Token discovery failure: token contains non-permitted character sequence (\\r
)

After investigation, he found out that:

  • Previous token (which corresponds to incorrect client details) is longer (contains more characters) than the correct one.
  • When client details are updated, refresh_token plugin writes new token "on top" of the old one, without truncating the file.
  • That's why garbage from the previous token appears on the second line of the file.

Adding the OS_TRUNC flag when opening token_out should solve that issue.

Thanks

Merge request reports