Not-Accessible remote share causes high slowdowns (and timeouts)

[krombel]

Steps to reproduce

1. add a remote share
2. shutdown or the server of the remote share (or let it respond 404)
3. experience HIGH loading times with even failing times

Expected behaviour

I just get an error that the share is not available anymore or similar

Actual behaviour

The access of (nearly) all pages becomes unusable slow (timeouts)
The client was not throwing any errors as it handles them silently I think

Nextcloud version: (see Nextcloud admin page)
12.0.4 and 13.0 beta 3 (updated to see if the update would resolve the issue)

Logs

Web server error log

from time to time I see

504 Gateway timeout

for the following requests:

/css/core/578c904904b0a852aad8a5b9acbd3044-share.css?v=97d63a6b-17
/js/core/merged-share-backend.js?v=97d63a6b-17 

Browser log

requests of
/ocs/v1.php/apps/files_sharing/api/v1/remote_shares?format=json&include_tags=true
are returning (after a long delay 2,5 minutes; only viewable after increasing timeouts)

{
    "ocs": {
        "meta": {
            "status": "failure",
            "statuscode": 998,
            "message": "Invalid query, please check the syntax. API specifications are here: http:\/\/www.freedesktop.org\/wiki\/Specifications\/open-collaboration-services. DEBUG OUTPUT:\n"
        },
        "data": []
    }
}

What have I done to resolve the issue

I could not list the shares on the webapp so I deleted that share from the database directly.

SELECT * FROM oc_share_external;
DELETE FROM oc_share_external WHERE id=<id of the share that is offline>;

After that the speed of the webpage is "back to normal"

[rullzer]

Fixed in master so should be in the next beta/rc (and thus NC 13): 3b343e2

[krombel]

Haven't found the issue for that...
So this is a duplicate of #7001

[rullzer]

yep, hopefully we can release an update soon in the new year. So you can test :)

[krombel]

The share that causes the issue will not reappear. So removing the share just fixed it.
Therefore this is not that critical and I cannot test it anymore (without spinning another instance up)
I opened the issue for "longterm-enhancement" :)
So thanks for having a fix implemented even before I requested it 😇