Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

Bug 320042

Summary:

improve handling of error Unknown code: 99

Product:

z_Archived

Reporter:

Steffen Pingel <steffen.pingel>

Component:

Mylyn

Assignee:

Frank Becker <eclipse>

Status:

RESOLVED FIXED

QA Contact:

Severity:

normal

Priority:

CC:

eclipse, robert.elves

Version:

unspecified

Target Milestone:

3.5

Hardware:

OS:

Linux

Whiteboard:

Attachments:

Description	Flags
screenshot	none
patch V1	none
mylyn/context/zip	none
patch	none
mylyn/context/zip	none
commited patch	none
mylyn/context/zip	none

Description Steffen Pingel

2010-07-15 17:06:51 EDT

Occasionally I get an error "Unknown code: 99" when submitting tasks to Eclipse.org (see screenshot). Usually submitting works if I retry. It would be nice if the error message was more descriptive or the connector would automatically retry submission.

Comment 1 Steffen Pingel

2010-07-15 17:07:06 EDT

Created attachment 174450 [details]
screenshot

Comment 2 Frank Becker

2010-08-16 14:52:08 EDT

Steffen,

you get this error in the following cases:

1) "expired_token"
2) "missing_token"
3) "invalid_token"

The problem is that you can only see in the body what problem exists with the token.

For me it looks like 1) is the most common error.
An token is valid for 3 days, so when you made change and do not sync for three days you can get this error.

Should we change BugzillaClient.parseRepositoryResponse

to add the following "Error with security token. Please retry operation." to the BugziallStatus

Or

do you think we need to parse the Responsebody for show the real reason?

Thoughts?

Comment 3 Steffen Pingel

2010-08-17 01:15:39 EDT

Thanks for clarifying. I would prefer if we could parse the body to provide a meaningful error and fall back to a standard error, e.g. "Security token error occurred during submission. Synchronize task and re-submit changes."

Comment 4 Frank Becker

2010-08-17 16:15:51 EDT

Created attachment 176842 [details]
patch V1

(In reply to comment #3)
> Thanks for clarifying. I would prefer if we could parse the body to provide a
> meaningful error and fall back to a standard error, e.g. "Security token error
> occurred during submission. Synchronize task and re-submit changes."



Patch commited to HEAD for the fall back case.

For passing the body I see the problem that we must find an test that work in all languages.

Thoughts?

Comment 5 Frank Becker

2010-08-17 16:15:53 EDT

Created attachment 176843 [details]
mylyn/context/zip

Comment 6 Steffen Pingel

2010-09-01 18:08:21 EDT

(In reply to comment #4)
> For passing the body I see the problem that we must find an test that work in
> all languages.

If that's not easily possible than let's not worry about it for now. We should file a but against Bugzilla to supply a unique ID for each error somewhere in the Http repsonse or HTML output (e.g. through a hidden field) so that we can identify each error reliably.

Comment 7 Frank Becker

2010-09-02 16:55:57 EDT

I open bug https://bugzilla.mozilla.org/show_bug.cgi?id=593138 for the request.

Comment 8 Frank Becker

2010-09-18 13:43:26 EDT

Created attachment 179178 [details]
patch

this patch is a save of my current implementation that I need when https://bugzilla.mozilla.org/show_bug.cgi?id=593138 is fixed

Comment 9 Frank Becker

2010-09-18 13:43:28 EDT

Created attachment 179179 [details]
mylyn/context/zip

Comment 10 Steffen Pingel

2010-10-27 08:08:36 EDT

Looks like the upstream bug has been fixed.

Comment 11 Frank Becker

2010-10-31 12:16:20 EDT

Created attachment 182113 [details]
commited patch

Comment 12 Frank Becker

2010-10-31 12:16:23 EDT

Created attachment 182114 [details]
mylyn/context/zip

Comment 13 Frank Becker

2010-10-31 12:18:17 EDT

Steffen,

can you please verify this.