atesin
commented
5 months ago more oddities found:
- ddupdate systemd unit file is enabled as a user service, i think would be less troubled to enabled as system wide service, is unlikely ddupdate would run on a multi-user environment but instead on a service network machine managed by a single sysadmin (root)
- ddupdate is configured to start at user login, that mean for example everytime user logs in for whatever reason, ddupdate is triggered even when nonsense, for example everytime user logs in local desktop environment, slowing down the process and introducing possible security issues (what have to do local user processes with internet dns ?????... in what depends??)
@leamas i can see your intention of trying to make ddupdate as much linux/posix standard possible, bringing ddupdate multi-user compatibility, trying to use system resources as keyrings .netrc files, sticking with file standards and formats, etc ... but in practice some of these features bring more complications than helps ... please try to be more practical and aware of sysadmin everyday real life than academic .... for example i see no harm in give ddupdate his own unique and exclusive config and security files and formats in /etc/ , in fact many system services have its own exclusive config files with its own format like smb.conf, my.cnf, httpd.conf, nginx.conf, etc ... and none of them use /etc/passwd file or /home/ directory for multiuser for example (except for process isolation), not even samba!
for example, i never had heard of .netrc file before (i found info that was used by ancient ftp clients), i worked flawlessly with network services (firewalls, services, etc) without feeling then need of the file, or even knowing its existence... as i only manage servers, root is the only account i use, i have no user accounts, so i never had to use the keyring ... access to all my servers are restricted and i have nightmare root passwords (unneeded tcp ports closed, and other security measures) ... i feel very comfortable on editing /etc/ conf files because most probably i am the only one that will enter the server .... but for this software, i had to install packages and learn to use files that i never had to
i think all these multiuser features are fine, but single user (root) should be the default and multi user should be optional... unless you have a vps hosting service, in which case i think is unlikely ddupdate was used for ddns updates but other more advanced tools instead like cpanel, plesk, etc
leamas
hi... i like the idea of this application, of being modular, unlike ddclient that i was lont time using it.. i have big hopes with this
however sometimes is a nightmare to configure... i recently moved to a debian server (raspberry pi actually), have to reinstall everything, and make DDNS to work is essential for me
however i had troubles from the beginning, until i like the approach of one configuration section for each service/account... i think it would be more practical/natural/flexible to have one configuration section for each DOMAIN
what happens if somebody has 2 domains in same ddns service?, or if someone updates some domain with some ip and its subdomain with another?, or if those domains in same ddns service have different user accounts (i.e. for different customers)?? ... that is what i had to write a plugin with non-standard
.netrcsyntaxso i migrated my both config files from my old fedora machine to this debian one (raspberrypy)... but a message appears saying my WHOLE
ddupdate.conffile is skipped because lack of [update] section :/ ... so nothing happens except the default 'dry-run' update... so i added the [update] label, but now just discards my other sections, etcspeaking about
.netrc... i never heard that file before, but according what i searched, is an ancient and poor documented file used to store credentials to ease REMOTE FTP connections, i guess nobody uses that file nowadays, for me it sounds like having a text file listing remote hosts (yes, thehostsfile, in times prior to DNS system)why NOT to make the same other applications do?, to have its OWN custom/exclusive/non-shared path/name/format file? ... for example, mysql have its own
my.cnffile, apache has its ownhttpd.conf, so nginx withnginx.conf(with different format), php hasphp.inifile that is exclusive and only read by php, samba hassmb.conffile, ssh has justsshd_config, and so onanother problem i faced were the
$XDG_CONFIG_HOMEand similar environment variables that threw me errors when i first randdupdate-config.... i found that those variables are related toX11/Xorg, but what happens if somebody has just text console with no graphical interface?, or even has a headless server with no monitor attached at all? (like me!)i was unable to make work ddupdate from debian package, i found is pretty outdated anyway and seems
$XDG-*vars were fixed, so i removed it and installed ddupdate frompipthen i was reading
man ddupdateabout files and configurations, and saw that ddupdate first reads$XDG_CONFIG_HOME/ddupdate.confif variable is defined (with visible file, prone to being accidentally modified/deleted), but if not it falls to~/.config/ddupdate.conf(should be if actual file exists instead if variable is defined)so ... what if somebody wants to use another custom path like
/opt/profiles/singleuser/ddupdate.configor/mnt/portable2go/ddupdate.ini? .... after all i think MOST of machines where ddupdate runs are single user boxes (routers, nas, embedded, home servers, small business, etc) .... (can-c <path>option be used in harcoded conf file, a lainclude?)ddupdate seem installed ok, man pages are displayed, etc... but when i run
ddupdate --helpthis happensthings start turn even weirder...
i made symbolic links in
/bin/, but now ddupdate floods my console with large and tangled python error messages... and ddupdate-config also complains me something about keyring package... dunnowhy the heck am i unable to run ddupdate fresh installed, and have to make ever more complicated manual researches and fixes everytime???
i am tired