Re: [PATCH] Reload SSL certificates on SIGHUP
От | Michael Paquier |
---|---|
Тема | Re: [PATCH] Reload SSL certificates on SIGHUP |
Дата | |
Msg-id | CAB7nPqQU50-UMuFcffmynDbH7_kBx3N-Q3ujL-nM2OF3X8biMA@mail.gmail.com обсуждение исходный текст |
Ответ на | Re: [PATCH] Reload SSL certificates on SIGHUP (Andreas Karlsson <andreas@proxel.se>) |
Ответы |
Re: [HACKERS] [PATCH] Reload SSL certificates on SIGHUP
|
Список | pgsql-hackers |
On Mon, Dec 5, 2016 at 9:50 AM, Andreas Karlsson <andreas@proxel.se> wrote: > On 12/04/2016 03:20 PM, Michael Paquier wrote: >> On Sun, Dec 4, 2016 at 11:11 PM, Andreas Karlsson <andreas@proxel.se> >> wrote: >>> On 12/04/2016 02:12 PM, Michael Paquier wrote: >>> Does this have a precedent in the code? >> >> >> data_checksums in guc.c is an example, it is marked with >> GUC_NOT_IN_SAMPLE | GUC_DISALLOW_IN_FILE and its value is updated with >> SetConfigOption() when the control file is read. The idea would be to >> do something like that with LoadedSSL. OK, here is attached what I had in mind as reload-ssl-v08-02.patch for reference. This applies on top of the main patch reload-ssl-v08-01.patch that is the same version as v7 with the issues I reported previously as addressed. LoadedSSL is mapped with a read-only GUC parameter that new sessions can query after connecting. The only use case where that would be useful would be when using sslmode=prefer to check whether the SSL context is loaded even if ssl has been switched from off to on. But let's be honest, pg_stat_ssl reports already this kind of information, making this patch at the end useless because LoadedSSL does not change for an already-spawned backend. > Thanks. I will be quite busy the upcoming couple of weeks so there will be a > while until I can sit down with this. Feel free to contribute to the patch. I am switching the patch as ready for committer. I have no more comments about this patch. Note to committer-san: please look only at v08-01. -- Michael
Вложения
В списке pgsql-hackers по дате отправления: