do you know how other do autocron ?

Question

when a cron request starts, it needs a requester not leaving the page. Else the process initiated is stopped.

The only way I found is to take the opportunity of a great download.
It's not realistic... if the downloader changes his mind and stops the download before the end of the task, the process is killed after some seconds

I have many solutions using smart servers , even if they don't cron explicitely but I'm still looking for a real independant autocron. I don't forget that 1.000.000 calls / day cost less than 10 $ / month ( small machine without apache, juste cron and mysql )

Splitting a cron task is rarely a good idea. Ie for dbsave, the database dumper, there is the risk to get a not coherent database

Do you know how other do ( without any root server trick , some webmasters are not sure to get their help ) ?

thank you for your tips

edit : I found a solution, where QA calls itself and both session use socket properties to get working the needed time

Comments

If a task is long, the only way I can think of is if it can be split into smaller tasks, so each page would do a little bit of it, and then go away.  Without going to root/exec, I don't see how this could be done otherwise. So would be interested to hear as well.

---

For the moment the best solution I found needs 2 QA installations that exchange files at ie 16 bytes / minute while they do their night tasks. It might be a solid option since the file exchange resets the time-out ( like a download, it can take hours ). Testing if it is time to cron doesn't add a great overhead ( if time is between limits and if file_exists(todo) ) ; it is not measurable in a big emulation test.

---

2 QA installations ? perhaps with 2 instances of the same site !!! yes !

Answer 1

Please check out WP backup https://wordpress.org/plugins/backupwordpress/
to see how they are doing it.  It should be possible to port the idea/code to Q2A.

"Works in low memory, "shared host" environments."

Comments

I want to write my own plugin , not to port the work of this guy :) Now, it's done and in test phasis ...

---

it works :) to get the cron done, one must expect a visit on the main page by anything even an engine. Overhead average : 0.75 ms when ordering the task and 0 otherwise. I still don't know how they do exactly but this solution handles only the dbsave cron ( as tonight )

---

it worked fine during 30 min several times. I'm glad because I saw pages and pages and didn't read any. I found all the models in the php.net documentation :) But , now I have to deal with the php memory limitations. The cron plugin is allmost ready.

---

memory limitations, done ... now, the mysql 'bit(1)' bug ... to get an universal test working , thinking to the abominable case of a new admin table with a 'bit' field.