Model contains an error: ValueError: Failed to process datasource.wikibase.revision.parent.item_doc: Expecting value: line 1 column 1 (char 0) Traceback (most recent call last): File "/srv/deployment/ores/venv/lib/python3.4/site-packages/revscoring/dependencies/functions.py", line 244, in _solve value = dependent(*args) File "/srv/deployment/ores/venv/lib/python3.4/site-packages/revscoring/dependencies/dependent.py", line 52, in __call__ return self.process(*args, **kwargs) File "/srv/deployment/ores/venv/lib/python3.4/site-packages/revscoring/features/wikibase/datasources/revision_oriented.py", line 108, in _process_item_doc return json.loads(text) File "/usr/lib/python3.4/json/__init__.py", line 318, in loads return _default_decoder.decode(s) File "/usr/lib/python3.4/json/decoder.py", line 343, in decode obj, end = self.raw_decode(s, idx=_w(s, 0).end()) File "/usr/lib/python3.4/json/decoder.py", line 361, in raw_decode raise ValueError(errmsg("Expecting value", s, err.value)) from None ValueError: Expecting value: line 1 column 1 (char 0)
With a closer look to the error. It's clear what's wrong. It occurs when the ORES extension tries to score edits not made in the main name space in Wikidata (ORES service doesn't support that and it's by design). The fastest solution is making the extension score only main namespace edits for Wikidata.
The patch here only enables this method, we can add as many namespaces as we want in mediawiki-config patches. Once this one is merged, I will make the config one and put both item and properties namespaces :)
Mostly because the models is meant to deal with wikibase data model not with the textual and wiki-like part. So when it tries to understand an edit that not made in main namespace (and ns:120, properties) it errors trying to parse the json since these pages are not json-parse-able.
It got deployed in June 20th, 23:40 UTC. After that time, we had 21 failed jobs again. Which my guess is that they are old jobs which got failed and they are retrying. For example yesterday, in the exact time span we had 71 failed jobs. I will keep monitoring to see if the failed jobs diminishes or not.
https://grafana-admin.wikimedia.org/dashboard/db/ores-extension shows this patch reduced failure rate of ORES extension jobs from 3.7% to 1.2%. Other reasons for failure is timeouts. I need investigate that