Hello sir,

There is a link to the changelog on the product’s page. Please follow all the steps so as not to break anything.

https://docs.php8developer.com/web-cronjobs/#/changelog?id=v-111

Regards,
Roman

Thanks. One more thing, I don’t know the version I have. Do i need to update version by version, or can I just upgrade to new version just following your steps ?

What the purchase date is?

Don’t remember, but I left my review 11 Months ago. That what it shows in the Reviews. I think in that same week. You help me configure I remember. Maybe on February 2023.

So you are on the penultimate version, so open the link I sent you, and carefully follow 6 steps.

Thanks, so I need to do V 4.14 all steps then V 4.15 ?

Can you re-open the link? I sent you the wrong at the first time. The version is 1.11

I re-open the link you send me. It has the full Log, from 4.0 to 4.15. I can see version by versions.

Penultimate version is V 4.14 ? (My reviews were around February 2023) and V 4.13 is from January 29, 2023.

I just need to upgrade: to V4.14 all steps and then V.4.15 ?

It’s hard to tell why without looking at the exact job. From the code point of view, each task emits a log line in any case. Can you provide more details?

1. Is it an HTTP or CLI task?

2. How do those “empty log” lines look like? Can you send me a screenshot by email?

1. This tasks is CLI

2. “empty log” is no log. For example, I noticed when I added new task, several times task lanches, but no any log added. In Statistic column i see “There are no statistics”, but task lanches and finishes corrects.

Currently, in our system 99 tasks in total. We have CLI and HTTP tasks, and periodically adding new tasks.

At one point, we also increased the limits in the configuration $batchUrl and $bathCommand, as we noticed that tasks simply did not start at the required time.

When you manually run that task (via the web interface) do you see an output? Does this task have the following ending or similar ” >/dev/null 2>&1”?

Increasing batch $batch* variables is a good idea when you have enough resources to start many processes simultaneously.

If I run it manually, it seems like everything is always fine.

Our tasks no any use the background execution suffix (...&).

I also notice that tasks that finish quickly and run frequently (for example, every minute or */2 minutes) may be delayed in starting.

For example, the timing of a task is set to executes every minute, and it takes 10-15 seconds to complete it. But the Cronjobs App shows that its next launch is 10:46:00, while the current time is already 10:52, and such a difference can reach, on average, up to 10 minutes.

We have a lot of tasks in the queue. Some work very quickly, some can work for several hours.

Native Linux-cron launches the master task of framework stably every minute – could there be some kind of collision or delay in the launch or in saving logs of completed tasks? This is a hypothesis, because I don’t know the exact logic behind how the Cronjobs works.

In terms of resources, there are enough of them – this is a dedicated server with several physical processors and a large amount of memory. Previously, all these tasks worked for us through native cron.

This may be the rub:

We have a lot of tasks in the queue. Some work very quickly, some can work for several hours.

Let me briefly explain how the master cron task works.

1. There are two types of jobs: CLI and HTTP. The jobs in these groups are executed one after the other synchronously.

2. Jobs in those groups are split into batches. This is where “batchUrl” and “batchCommand” emerge.

3. Let’s say “batchCommand” is equal to 10, but we have 15 jobs in the queue. The script gets the first 10, runs all of them asynchronously and waits until all of them are finished. The remaining 5 tasks are waiting for the next batch to process – they are not picked up by the new process of master script.

4. The next time launch calculates just before batch processing. This is done because some tasks may run slowly, so another instance of the master PHP script, which is triggered by the native Linux cron job system, will pick up those jobs for processing, even when the slow batch group is still running in the previous master script.

So, how to address the situation when you have a lot of fast and slow scripts?

Well, you can increase the “batchCommand” to the total number (or more) of your jobs. So all of them would be processed in one large queue asynchronously (make sure, you don’t have limits of the maximum child processes).

Alternatively, you can try to install another instance of web-cronjob into a subdomain, set up the master PHP cron script for that instance, and run there only slow tasks.

Speaking about the issue with logs: try to increase the wait_timeout directive in the my.cnf MySQL configuration file and restart the database.

wait_timeout = 28800

Regards,
Roman

Thanks for descriptions above.

Regarding to logs and timeout – this is also quite possible. But we specifically set a low global timeout in the server settings, because there is one DBsrv, but there are many applications.

And we have moved the timeout change to the application level through the request “SET wait_timeout=...” after establishing a connection. Each our app and script decides for itself how much processing time it required.

Is it possible in the Cronjobs somewhere in the config to set a higher timeout when using DB, or let app check connection before try save logs (or any manipulation with DB), and if no connection, perform reconnect.

Is it possible in the Cronjobs somewhere in the config to set a higher timeout when using DB

Yes, you can do that. Try to open the protected/config/db.php file and append the following lines to the array:

<?php
return [
...

    'on afterOpen' => function($event) {
        $event->sender->pdo->exec('SET wait_timeout=28800');
    },
];

Didn’t test in the wild, but should do the trick.

Hey. Can you send me FTP credentials? I’ll take a look what is wrong with your script.

I’ve sent you the ftp access.

Hello sir,

Can you contact me by email? Include please FTP credentials.

Regards, Roman

can you do remote with anydesk software? I will send anydesk id via email

Let’s try, but we need to coordinate the time.

Please send me cPanel / plesk / etc credentials by the email.

all ok, thank you so much