Re: Remove non-fast promotion Re: Should we remove a fallbackpromotion? take 2
От | Fujii Masao |
---|---|
Тема | Re: Remove non-fast promotion Re: Should we remove a fallbackpromotion? take 2 |
Дата | |
Msg-id | 1c177a61-e24e-461a-e046-eca80e9ee0fb@oss.nttdata.com обсуждение исходный текст |
Ответ на | Re: Remove non-fast promotion Re: Should we remove a fallbackpromotion? take 2 (Alvaro Herrera <alvherre@2ndquadrant.com>) |
Список | pgsql-hackers |
On 2020/04/22 6:57, Alvaro Herrera wrote: > On 2020-Apr-20, Fujii Masao wrote: > >> + /* >> + * In 9.1 and 9.2 the postmaster unlinked the promote file inside the >> + * signal handler. It now leaves the file in place and lets the >> + * Startup process do the unlink. >> + */ >> + if (IsPromoteSignaled() && stat(PROMOTE_SIGNAL_FILE, &stat_buf) == 0) >> { >> - /* >> - * In 9.1 and 9.2 the postmaster unlinked the promote file inside the >> - * signal handler. It now leaves the file in place and lets the >> - * Startup process do the unlink. This allows Startup to know whether >> - * it should create a full checkpoint before starting up (fallback >> - * mode). Fast promotion takes precedence. >> - */ > > It seems pointless to leave a very old comment that documents what the > code no longer does. I thikn it would be better to reword it indicating > what the code does do, ie. something like "Leave the signal file in > place; it will be removed by the startup process when ..." Agreed. And, while reading the related code, I thought that it's more proper to place this comment in CheckPromoteSignal() rather than CheckForStandbyTrigger(). Because CheckPromoteSignal() actually does what the comment says, i.e., leaves the promote signal file in place and lets the startup process do the unlink. Attached is the updated version of the patch. Regards, -- Fujii Masao Advanced Computing Technology Center Research and Development Headquarters NTT DATA CORPORATION
Вложения
В списке pgsql-hackers по дате отправления: