Page MenuHomePhabricator

Internal error: Could not determine if the copy succeeded.
Open, HighPublic

Description

Since a few days I get reports, incl. from myself as Wikimedia Commons uploader, like "Internal error: Could not determine if the copy succeeded." after click to submit the descriptions.

Clicking on "Retry failed uploads" I get
"[api-error-internal_api_error_UploadStashFileNotFoundException]"

For my uploads the file uploads were completed anyway but I hear about half done uploads (file page without file).


Version: 1.22.0
Severity: major
See Also:
https://bugzilla.wikimedia.org/show_bug.cgi?id=36587

Details

Reference
bz43967

Event Timeline

bzimport raised the priority of this task from to High.Nov 22 2014, 1:14 AM
bzimport set Reference to bz43967.

Reproduced on James_F's laptop, but the files actually completed.

Is this fixed now? Or are there any steps to reproduce this?

*** Bug 48647 has been marked as a duplicate of this bug. ***

looks like chunked upload issue

For the record, since I had to grep the feedback page for it, the Russian comment dating September 24th 2013 just states that they've encountered the error, nothing more.

I confirm that this is probably coming from chunked upload-specific APIs. More specifically "checkStatus", which allows to check when an upload has been reassembled when the chunk API keeps replying "Poll".

It seems like something almost identical is also exposed to API consumers through "statuskey", but UploadWizard doesn't seem to make use of it.

The failure is due to the session entry for the requested filekey being missing. I will investigate further once I have shell access approved to access logs. Something I'm particularly interested in finding out is whether all the session data is missing or just the entry for the particular file we're interested in.

Just got this on https://commons.wikimedia.org/wiki/File:Practica_D._magistri_V00241_00000004.tif

Error console says the following, dunno if related:

2 Validation error against schema UploadWizardFlowEvent: Unrecognized property: quantity
8 Failed to load resource: the server responded with a status of 503 (Service Unavailable) https://commons.wikimedia.org/wiki/Special:UploadStash/thumb/12mniz1s4dlw.l7rnkh.4135183.tif/lossy-page1-78px-12mniz1s4dlw.l7rnkh.4135183.tif.jpg

(In reply to Gilles Dubuc from comment #9)

I confirm that this is probably coming from chunked upload-specific APIs.
More specifically "checkStatus", which allows to check when an upload has
been reassembled when the chunk API keeps replying "Poll".

It seems like something almost identical is also exposed to API consumers
through "statuskey", but UploadWizard doesn't seem to make use of it.

The failure is due to the session entry for the requested filekey being
missing. I will investigate further once I have shell access approved to
access logs. Something I'm particularly interested in finding out is whether
all the session data is missing or just the entry for the particular file
we're interested in.

Is this a sudden increase? If so, maybe hhvm related? chunked upload does some scary stuff with sessions, it wouldn't shock me if hhvm interfered with that somehow.

I'm adding HHVM to the keywords, just in case.

Is this still a problem/still happening? If not, maybe this tasks priority could be lowered or declined or something?

Is this still a problem/still happening? If not, maybe this tasks priority could be lowered or declined or something?

There is no specific reason to believe something may have fixed this since December, AFAIK. I don't know if errors are still being logged somewhere.