[feature request] Batch API more info on failed/error domains

[rudiedirkx]

Currently the Batch API doesn't give any details on why a domain 'failed'. It just says status = error, and nothing else. The web UI status requests give more information during testing, and sometimes (very rarely) when the test fails, it tells you which category failed. Not a lot of information, but at least it's some. Can the Batch API provide some more info on failed domains too?

[rudiedirkx]

Web UI status JSON that I'm talking about (for mail):

[
    {
        "name": "ipv6",
        "done": true,
        "success": true
    },
    {
        "name": "dnssec",
        "done": true,
        "success": true
    },
    {
        "name": "auth",
        "done": true,
        "success": true
    },
    {
        "name": "tls",
        "done": false,
        "success": null
    },
    {
        "name": "rpki",
        "done": true,
        "success": true
    }
]

Sometimes (VERY rarely) there's a done = true but success = false. That's interesting to know in the Batch API too. Or whatever else could be the reason for a failing Batch API domain. They happen more often in the Batch API (max 0.1 %, but that's way more), so maybe there's another reason.

[bwbroersma]

Instead of /results, try /results_technical.
Note that there are some known (open) issues with the current /results_technical.
See the OpenAPI specifications.

[rudiedirkx]

Instead of /results, try /results_technical.

Ooeeh interesting. Is there a way to trigger an error domain so I can dev this into my script? 😆 There are too few errors. What a problem to have.

How long are results stored after "done"? I want to try the next "error" manually so I can see what kind of technical details you have. A request from 9 Sep was already unavailable yesterday, so it can't be long.

[bwbroersma]

/results_technical also include data when there are no errors.

• I'm unsure how long the JSON results are cached Update: the results are cached for >6 days, but at the moment longer (Implement delete batch results #1509), however they are stored in the DB and can be re-generated on request.

Calling /request/id will trigger the "status": "generating", if it's "status": "done" you can fetch /results and /results_technical.
A few generating report requests which where queued during the clogged queue are now in limbo (since the queue was cleared), we're trying to recover from that, but it will also recover in a while (because of this cache).

[rudiedirkx]

JSON for failing domain in /results_technical:

    "www.herbestemmingnoord.nl": {
        "status": "error"
    },

Well, that's not useful 😆 I don't need the technical results for all the passing domains, I'm looking for some detail on why the failed failed.

[rudiedirkx]

Some domains keep failing in the Batch API, and I can't see any details on why. Is there any way to find out? results_technical doesn't include any error info.

Those domains (I know of www.herbestemmingnoord.nl and www.werkenbijgroningen.nl) do sometimes work in the web interface, just not in the Batch API, so I'm curious for more info on API errors.

[WKobes]

For www.werkenbijgroningen.nl: see DigitalTrustCenter/sectxt#79 which will work again when we update the sectxt package in our next release.

[rudiedirkx]

https://www.werkenbijgroningen.nl/ still breaks every time. mail works, inet fails. No error details.

[WKobes]

https://www.werkenbijgroningen.nl/ still breaks every time. mail works, inet fails. No error details.

This is fixed in the latest sectxt release which will be included in v1.10 (https://github.com/internetstandards/Internet.nl/blob/main/Changelog.md#110-in-progress).

[rudiedirkx]

https://www.werkenbijgroningen.nl/ worked for a while, but now it breaks again. I can't see any error details (which is what this issue was about), so I don't know if it's the same thing as before.

---

Batch (1.9.3?) is still broken, web UI (1.10.0) still works after the sectxt upgrade. So nothing new to report really. Sorry. Looking forward to the 1.10 upgrade in Batch.