Tout dépend de ce que l'on veut. Ils ont répondu à ces critiques dans la FaQ de openntp :
6.12.1 - "But OpenNTPD isn't as accurate as the ntp.org daemon!"
That may be true. That wasn't OpenNTPD's design goal, it was intended to be free, simple, reliable and secure. If you really need microsecond precision more than the benefits of OpenNTPD, feel free to use ntp.org's ntpd, as it will remain available through ports and packages. There is no plan or desire to have OpenNTPD bloated with every imaginable feature.
6.12.2 - "Someone has claimed that OpenNTPD is 'harmful'!"
That would probably be Brad Knowles. If accurate time keeping is important, a number of users have reported better results from OpenNTPD than from ntp.org's ntpd. If security is important, OpenNTPD's code is much more readable (and thus, auditable) and was written using native OpenBSD function calls like strlcpy, rather than more portable functions like strcpy, and written to be secure from the beginning, not "made secure later". If having as many people using time synchronization is valuable, OpenNTPD makes it much easier for larger numbers of people to use it. If this is "harmful", we are all for it.
There are almost certainly applications where the ntp.org ntpd is more appropriate, however it is felt that for the other 95% of the users, OpenNTPD is more than sufficient.
Bref 2 outils différents pour 2 besoins différents. J'ai lus les critiques de Brad Knowles, et je vois peu de problèmes critiques. Son principal délire, c'est qu'il faudrait avoir l'heure à la nano seconde près dans les logs au cas ou on devrait donner des preuves d'une intrusion devant un tribunal.
[^] # Re: openntp
Posté par Psychofox (Mastodon) . En réponse au journal Fork de CVS. Évalué à 2.
6.12.1 - "But OpenNTPD isn't as accurate as the ntp.org daemon!"
That may be true. That wasn't OpenNTPD's design goal, it was intended to be free, simple, reliable and secure. If you really need microsecond precision more than the benefits of OpenNTPD, feel free to use ntp.org's ntpd, as it will remain available through ports and packages. There is no plan or desire to have OpenNTPD bloated with every imaginable feature.
6.12.2 - "Someone has claimed that OpenNTPD is 'harmful'!"
That would probably be Brad Knowles. If accurate time keeping is important, a number of users have reported better results from OpenNTPD than from ntp.org's ntpd. If security is important, OpenNTPD's code is much more readable (and thus, auditable) and was written using native OpenBSD function calls like strlcpy, rather than more portable functions like strcpy, and written to be secure from the beginning, not "made secure later". If having as many people using time synchronization is valuable, OpenNTPD makes it much easier for larger numbers of people to use it. If this is "harmful", we are all for it.
There are almost certainly applications where the ntp.org ntpd is more appropriate, however it is felt that for the other 95% of the users, OpenNTPD is more than sufficient.
Bref 2 outils différents pour 2 besoins différents. J'ai lus les critiques de Brad Knowles, et je vois peu de problèmes critiques. Son principal délire, c'est qu'il faudrait avoir l'heure à la nano seconde près dans les logs au cas ou on devrait donner des preuves d'une intrusion devant un tribunal.