Re: [GENERAL] How to handle simultaneous FOR-IN UPDATE-RETURNING loops?

Поиск
Список
Период
Сортировка
От David G. Johnston
Тема Re: [GENERAL] How to handle simultaneous FOR-IN UPDATE-RETURNING loops?
Дата
Msg-id CAKFQuwb4EC8e4DwA0J5qujUUHJ5kNd-6Mp3DESPqVDOFCibGVQ@mail.gmail.com
обсуждение исходный текст
Ответ на [GENERAL] How to handle simultaneous FOR-IN UPDATE-RETURNING loops?  (Alexander Farber <alexander.farber@gmail.com>)
Ответы Re: [GENERAL] How to handle simultaneous FOR-IN UPDATE-RETURNING loops?
Список pgsql-general
Дерево обсуждения
On Mon, Jul 10, 2017 at 7:32 AM, Alexander Farber <alexander.farber@gmail.com> wrote:

However there is a problem: I can not use a "single-instance" cronjob to run words_expire_games hourly.

​Setup a cron job that invokes the servlet - probably via "curl"

My question is if I should ensure that only 1 servlet runs the custom PL/pgSQL function by using "synchronized" in Java as I do it right now:

​Probably not.  UPDATE takes out a lock that will prevent other updates from acting on the same records concurrently.​
 
Or if maybe there is some condition (maybe "UPDATE SKIP LOCKED"?) I could add to my custom function copy-pasted below? -

Why are you trying random syntax that isn't documented?

https://www.postgresql.org/docs/9.6/static/sql-update.html​

                UPDATE  words_games
                SET     finished = CURRENT_TIMESTAMP
                WHERE   finished IS NULL

​That should be sufficient.  Do you have any examples that show it is not?

In short, one of the main reasons for "UPDATE RETURNING" is so that one needn't determine the records to be updated separately from the actual act of updating.  Instead you update first and then capture the results for subsequent use.

David J.

В списке pgsql-general по дате отправления:

Предыдущее
От: Brian Dunavant
Дата:
Сообщение: Re: [GENERAL] How to handle simultaneous FOR-IN UPDATE-RETURNING loops?
Следующее
От: Tom Lane
Дата:
Сообщение: Re: [GENERAL] Changing collate & ctype for an existing database