[Linux] Rarely, .xlcore/patch will cause transient "Updater returned no integrity" errors

[FinishedFragment]

This has been happening to a good amount of users including yours truly. This last happened to me just two days ago but fortunately I found a solution this time, https://discord.com/channels/581875019861328007/850387047927250944/1370811151494942842, which involves deleting ~/.xlcore/patch.

So far, this has happened to me around 5 times. This would go away after an hour or two, sometimes in even less time. The disk is fine and healthy. It is a nvme SSD with ext4, though this should be irrelevant. This does not appear to depend on Dalamud being loaded, as I got this error without it being loaded in the past.

I was advised to try the "repair game files" option but all of the files are perfectly valid. I take care to shut down the game cleanly so that should also not be the cause.

I am certain this will happen again and I would like to debug why this is happening. I can neither control nor anticipate when this will happen again and this does not exactly seem to correlate with patches happening, as the patches were days if not weeks old.

Please tell me what I can do to help track down why this happens when this happens again.

Environment:

• OS: Arch Linux (vanilla, not some derivative)
• Version: 1.2.1.r19.ge8f1f22-1 (git, via AUR)

[FinishedFragment]

This happened again and:

---> System.Net.Http.HttpRequestException: Resource temporarily unavailable (kamori.goats.dev:443)

I think the error message "Updater returned no integrity" is just very misleading here. Something better would be "Our proxy is temporarily overloaded. Please try again later.".
Now I am unsure if the other cases of the error message are the same cause. I will observe this some more and see if I get anything other the server issue there, it is possible the .xlcore/patch thing is a red herring, but I can not say at this stage.

[FinishedFragment]

After some more time and debugging I think I found the weak point: DNS. The .xlcore/patch directory might have been a red herring because that meant I closed xivlauncher and started it up again.

My PC starts up very fast because I have a nvme ssd. When I log into my user account and start xivlauncher, it is possible that my locally run DNS server is still starting up and thus can not yield name resolution until another 500 ms or something has passed, especially since I am using dns over tls so any dns reply will take longer by design due to the full tls handshake happening, although it is caching replies after and also prefetches things that are in demand so this is effectively negated.

xivlauncher does not appear to retry name resolution when it failed once or it may only wait for 50 ms for an answer and then aborting. I have to actually close it and open it up again to make it retry. Literally no other software I use has this issue.

However, I can also not get it to spit out more details than "Resource temporarily unavailable" so I will need some instructions how to increase verbosity there as I am just guessing here. It sure sounds likely but I will need to know for sure.