130 Over Capacity -- Update still posts

[strawberry0948]

I’m occasionally seeing 130 over capacity errors when creating status updates through the API. This in itself isn’t a problem but at times the status updates will go through even though the api responds with the 130 error. Is this a known issue?

[andypiper]

I’ve not heard other reports of this - is this recent? Are you able to narrow down a particular scenario where this is occurring?

[strawberry0948]

I’m able to reproduce this when creating an update with the following image:

https://s3-us-west-2.amazonaws.com/edgar-images-production/users%2F15107%2F2twc2tdnugnrw2g5-users%252F15824%252Fzfm25roh71f-P1120809.jpg

Sometimes we see an over capacity error (but the update still appears on the feed), sometimes the request times out, and sometimes it succeeds with no problems.

[DooohWork]

Hi there, I have seen this. Our call of the Twitter API yields an http status of 503 and the response body is:
{“errors”:[{“message”:“Over capacity”,“code”:130}]}

but yet it appears that the post gets sent out anyways. From our perspective we look for and attempt to handle such errors by retrying but then we notice that the post was sent out twice. It appears Twitter still sends out the attempt that is reported back to us as having failed.

Is this a known Twitter bug?

[andypiper]

Is this via the statuses/update endpoint, or another one?

[DooohWork]

Yes

[andypiper]

It’s not a known current issue, but the over capacity error can occur during spikes of activity on the platform. I’m a little surprised that you’re then able to post an identical / duplicate status, as I would expect error code 187.

[DooohWork]

Well, we of course do try to handle these errors by retrying but we are expecting that if we are told of such an error that it is ok for us at that time, to retry but seeing that Twitter retries such errors themselves, internally(?) kind of makes the error useless, doesn’t it?
Oh and yes, I would have expected to also get a duplicate post error generated on the retry but we did not, although there was about 30 seconds or so between the initial failed attempt, which then caused us to retry and the successful retry.