Role change not occurring at EOT

Leaves the TinyMCE and Google Analytics Dashboard – but I have troubles believing it’s one of them.

Me too. That’s why I remain curious as to what are the plugins similar to yours that apothecarytravels uses.

What extremecarver uses on the left, what I use on the right:

Similar plugins:

Responsive Menu —— Privileged Menu
Autoptimize —— Optimize Database
Velvet Blues Update URLs ——- Go Live Update URLS

fyi: autoptimize does not do db optimization, but JS/ CSS aggregation & minification.

frank (ao dev)

Actually, only the last couple are genuinely analogous to each other. If I were you two, I’d be watching what happens with Velvet Blues Update URLs and Go Live Update URLs. FYI, I myself had some bad experiences with Velvet Blues Update URLs and so stopped using it some time ago, but I’ve never used Go Live Update URLs.

Thanks guys. Good info.

About Velvet – I did use it to update my site to https only – worked a treat there (much better than other plugins). The only other use is that if I see broken links google webmaster tools – I sometimes update them using Velvet. I could actually deactivate it and only activate when I need it. Again pretty sure it’s not the problem here.

Autoptimize I use together with a caching plugin – it packs most css/jss into the footer. It’s the best solution to achieve high scores on google speed tests. Actually with PHP7 and http/2 it should not be needed anymore. But so far neither Apache / nor Nginx support http/2 push – so I will keep it for some more time. (More than any plugin/other optimization my website was speeded up by moving to PHP7. Nearly as drastic as load decrease from nginx vs apache or any cache plugin vs none).

Right now it deletes itself about every 2 days on one of my websites. On the other one it’s fine. Really dunno what’s happening here.

It seems to be a generic WP problem. See this thread.

Have you both checked the amount of memory you have available?

Well I had PHP.ini set to 512MB – same for wp-config
— So now I’ll try
PHP 1.2GB
big website 768MB, small website 256MB (here the job was rarely lost).

I’ll have to read in how to setup automatic rescheduling – My big website got 10K users with subscription – maybe the script is super memory hungry run against 20k users, of which 10k have subscription?
Really s2member should periodically check itself to reschedule – this problem is not that uncommon.

However it’s the only cron job that I’ve ever lost (and there are about 20 in my cron events list) – so the error is clearly on the s2member side (maybe the job is not well written for big user database?).

It would not be the only place where s2member is coping badly with large amount of users.

Some more findings about it:
The only solution to it breaking down regulary is to increase the time it runs. So instead of running every 10minutes – I changed it to run only every 6 hours once. That helped a lot. I only lost it once since then (within 4 weeks). With 10min interval It rarely survived a day.

Changing memory did not help at all. At least with 512MB or more – it’s fine.

Still wonder why I’ve never lost any other cron job. I’m really sure it’s not a generic WP problem – but a s2member specific problem.

Hi extremecarver,

Thanks for this. So, are you saying that it’s better to change the “Recurrence” to every 6 hours?

Thanks.

yes – if 10 demotions every 6 hours are enough for your site – go for it. (or is it 20 per run?).

Ultimately s2member really should check if the cron job is lost – and reenable it. Anyone got a good idea how to create a cron job that runs every 24 hours and checks if the s2member cron job is still active – and if not re-enable it?

I think that “next run” of the cronjob kills the one, which was not finished yet. Means, you go to your office/club/home etc and see your colleague/friend/spouse counts flies (silly example, huh), and say “you go home/drink/sleep, I will finish counting”. They (old cronjob) stop his count, and you (new cronjob) starts (instead to continue) counting again. But in need time you can not finish, so next time when your colleague/friend/spouse comes, you (now old cronjob) are not finished, but they (now new cronjob) say “go home/drink/sleep, I will finish counting”, and starts over again. And this way the job is never finished… I am seen such things before. Not sure if the problem is same, must dig in s2M code, someone of developers may explain better.