[2.1 & 3.0]: Consider disabling cron during patches, mod installs, etc, when source code is in flux

[sbulen]

Basic Information

It may make sense to disable cron during patching. During patch testing, I occasionally see cron errors. If you do enough patches, it'll happen...

2 examples during a recent test are on the forum test board.

I just got another one now. When these ones come, they come in bursts of ~11 (which I think are the unicode updates):

Steps to reproduce

1. Intermittent issue, when cron kicks off while doing patch tests

Expected result

No errors

Actual result

Cron errors as listed above

Version/Git revision

2.1 (but can happen in any)

Database Engine

All

Database Version

8.4.4

PHP Version

8.4.5

Logs

Additional Information

No response

[jdarwood007]

Shouldn't be overly difficult. If $upgradeData from Settings.php is non-empty we have an upgrade in process.

[sbulen]

Shouldn't be overly difficult. If $upgradeData from Settings.php is non-empty we have an upgrade in process.

I'd be concerned about any time code is in flux: patch installs, mod installs, upgrades, etc.

I've done so many patch tests, I've seen a LOT of cron errors while doing so lately...

[jdarwood007]

Yeah, so this occurred when you started upgrade.php, but not finished? This means that $upgradeData in Settings.php is a non-null value.

Something like this in cron.php:

// Don't do john didley if the forum's been shut down completely.
if ($maintenance == 2 || !empty($upgradeData))
	die($mmessage);

[sbulen]

No - this happened installing the 2.1.5 patch.

Many times....

[Sesquipedalian]

#8703 should fix this for 2.1.

#8652 includes a version of the same changes for 3.0.

[sbulen]

Well, new info...

I just got the 11 errors on my production forum (2.1.6, unix, php8.4.5). 1 minute after my Weekly Maintenance ran, according to the scheduled task log.

No conflict with patching - or any other - activity this time. Just weekly maint.

I had applied 2.1.5/2.1.6 earlier this week.

[Sesquipedalian]

In that case, the issue occurring during the Unicode update is unrelated to patching. I will look into it further.