To get or not to get (app updates): App store load mitigation, app store hickup, or my config?

ℹ️ Support
1

AFTER I JUST updated all my NC25 RC5 instances using php occ app:update --all,
I issued this command for them and was confused about the outcome:

$ php occ app:list|grep " mail:";php occ app:update --all

domain1
  - mail: 2.0.2

domain2
  - mail: 2.0.2
related_resources new version available: 1.0.1
related_resources updated

domain3
  - mail: 2.0.1
related_resources new version available: 1.0.1
related_resources updated
spreed new version available: 15.0.0
spreed couldn't be updated
richdocuments new version available: 7.0.0
richdocuments couldn't be updated
mail new version available: 2.0.2
mail couldn't be updated

Questions are:

  • Why doesn’t domain3 see/get the mail update?
  • Why doesn’t domain1 see/get the related_resources update?
  • Why does domain3 sees/gets these updates only now? (All others had them already)
  • Why couldn’t spreed, richdocuments and mail not be updated on domain3 (while they had been updated fine for the other domains?)

In a consecutive runs of the occ app:update --all, some apps were updated for domain3:

spreed new version available: 15.0.0
spreed updated
richdocuments new version available: 7.0.0
richdocuments couldn't be updated
mail new version available: 2.0.2
mail couldn't be updated

richdocuments new version available: 7.0.0
richdocuments updated
mail new version available: 2.0.2
mail couldn't be updated

Really weird (=an error?!) is that upon next issue of occ app:update --all on domain3 yields “” (all apps up to date), which is not true as mail still is 2.0.1:

occ app:update --all
occ app:list|grep " mail:"
  - mail: 2.0.1

Notes: I did all this because I use scripts to keep my instances up to date. And checking their output I was confused that some got some app updates while others did not. As the above instances all run on one machine in one NGINX as virtual servers and as even their systemd.timers are mostly identical, that seems to be an issue with the NC store. I could be on purpose: Is there some load mitigation? Any insight anyone?

Seems irrelevant, but anyway: Arch Linux, NGINX, PHP 8.1, Mariadb, Redis

Is this the first time you’ve seen this error? No.

Steps to replicate it: Update several similar configured instances at once.