problem during authentication

[thpar]

While trying to migrate issues from a GitHub Enterprise to github.com, I run into

ERROR: There was a problem during authentication.
Double check that your username and password are correct, and that you have permission to read from or write to the specified repositories.

Weird enough, the milestones, labels, and the first 11 issues (with comments) are succesfully migrated, so I don't understand where the sudden 401 is coming from.

[thpar]

This was fixed by adding a delay in between issues. I used time.sleep(10).
Apparently github.com doesn't appreciate it when you send too many requests in a short time.

[thpar]

Too early to close this.
After about 50 issues (using the delay 10s) I get the same authentication error again.
Is noone else getting this behaviour?
Would it help to not use a delay, but catch this exception and then waiting for an amount of time and trying again, instead of simply exiting the script?

[thpar]

I tried the catch and retry approach and managed to import some more issues, until not a single request could come through any more. My guess is my IP is now flagged as spam.

Has anyone ever tried to migrate a large (250) batch of issues to github.com and succeeded?
Doing the same thing on our private Enterprise Github didn't pose any problems and happily migrated all our issues.

[kewinrausch]

Had the same issue. I wrote a new script for myself looking at this page: https://developer.github.com/v3/#rate-limiting

I you have your script certificated(Settings > Applications > Personal access tokens ) then you can do 20 requests per minute, otherwise way less.

Now I'm currently moving something like 450 issues, and it will take about 30 minutes(sleep every 5 seconds).

[kewinrausch]

New report:
sometimes, even after respecting the limits, github kick you off. Waiting some time and try again allows you to continue with the procedure(maybe renewing the Access token allows you not to wait; i have to test it now).

In case of error the script is useless, because there's noting which allows you to skip already inserted entries. I modified it to inplement such functionality(and of course skip creation of milestone and labels if there's something to skip) and it's working fine.

This is a functionality which definitely have to bo included in the official script(like --skip ). Skipping entries(milestones and labels) does not anymore terminate the script with the "422 Unprocessable entry" error.

EDIT: Using the script is also possible to close the issues and preserve the assignee(I did it). This is a functionality which can be useful too to add to this script.

[brianwc]

@kewinrausch Could you post your changes or even issue a pull request here? I'm experiencing the same problem and sleeping for 4 seconds between each request does not seem to be good enough. Seems like a conditional that would check the user's rate-limit and confirm it's >0 before proceeding would be better. Interested to see what you came up with.

[kewinrausch]

@brianwc I didn't fork this project, so I have only a local copy of the modifications.

I renamed the .py main file as .png and posted it at the end of this answer; just save it on your pc and open it with a text editor(or rename the extension as .txt or .py). At the begin you'll find some global variables I used to set the wait time between each request, after how many issue stop, comment to skip(if the process stop while processing a issue comment), issues to skip(already done in the previous session) and issue+pull request displacement.

The last variable is needed if you import the issues in a repo. which already contains other issues.
Take in account to not add any issue to the source repo while doing the import, otherwise you can break the '#' link used to refer between issues.
Every imported issue has the "imported" label.

Feel free to ask me for additional info, and perform some test before using it on the real repo.

[seannybgoode]

Any luck getting this to work? We have like 400 issues, I don't want to do it by hand.

[kewinrausch]

I already exported 400+ issues and pull requests in groups of about 100 per time.

Use my gh-issue-import image (download and rename as a .py file) and read my previous post.

[seannybgoode]

I ended up doing it ~5 issues at a time. Took me like an hour and my business partner got like 400 emails, but it's done now
.

[kewinrausch]

They can unsubscribe to events during the import procedure, so no spam in that case.

I had to do 100 and wait for 30 minutes, usually, to respect the limits. The only faster way is to build and register an application.

[seannybgoode]

If it was more than an occasional problem I probably would.

[nmccready]

Anyway the auth can be changed to use the .netrc file then all of this mess can be avoided?

[landonreed]

I've just created a PR that contains the changes mentioned by @kewinrausch in #41 (comment). Here's a link to the file changes if anyone that stumbles on this thread is interested.