Malone swallows comments

Bug #30823 reported by Andreas Schildbach
4
Affects Status Importance Assigned to Milestone
Launchpad itself
Triaged
High
Stuart Bishop

Bug Description

I just entered a comment on bug 29881 (an Ubuntu bug). This comment was swallowed and does not appear in the bug. I re-entered the comment with a slightly different wording two minutes later. This time it worked.

I received two confimation e-mail, one for each comment. Here are the message id's:

<email address hidden> (this is the one that was swallowed)
Date: Wed, 08 Feb 2006 12:45:17 -0000

<email address hidden> (this one was ok)
Date: Wed, 08 Feb 2006 12:47:06 -0000

Tags: lp-bugs
Revision history for this message
Björn Tillenius (bjornt) wrote : Re: [Bug 30823] Malone swallows comments

On Wed, Feb 08, 2006 at 01:01:13PM -0000, Andreas Schildbach wrote:
> I just entered a comment on bug 29881 (an Ubuntu bug). This comment was
> swallowed and does not appear in the bug. I re-entered the comment with
> a slightly different wording two minutes later. This time it worked.

After entering the first comment, what happened? Did you get an error
message, or did it say the comment was added, or did you simply return
to the bug page with no notification message at all?

Revision history for this message
Andreas Schildbach (schildbach) wrote :

As far as I can remember, there was no noticable error. At least there was not the well-known error-page with a ticket code, as it happens from time to time on Launchpad.

I think I was just returned to the bug page, as expected. Only the comment was not there.

Revision history for this message
Brad Bollenbach (bradb) wrote :

On 8-Feb-06, at 8:41 AM, Andreas Schildbach wrote:

> Public bug report changed:
> https://launchpad.net/malone/bugs/30823
>
> Comment:
> As far as I can remember, there was no noticable error. At least there
> was not the well-known error-page with a ticket code, as it happens
> from
> time to time on Launchpad.
>
> I think I was just returned to the bug page, as expected. Only the
> comment was not there.

Was the comment made from the +bug/NNNN page, or the +editstatus page?

Can you reproduce this problem reliably? Has it happened before or
since?

Cheers,

--
Brad Bollenbach

Revision history for this message
Björn Tillenius (bjornt) wrote :

Stuart, can you see if you find what the problem is? I can find both comment additions in the web logs, no errors, but the given message id doesn't seem to be in the database.

Changed in malone:
assignee: nobody → stub
status: Unconfirmed → Confirmed
Revision history for this message
Andreas Schildbach (schildbach) wrote :

Both comments were made from the normal bug page (+bug/NNNN), I did not (try to) change any status. I cannot reproduce the problem, I have written many comments before and since, but it only happened once.

Revision history for this message
Stuart Bishop (stub) wrote :

I've got nothing in the logs or database that helps me track this down. The email should not have been sent if the requet failed in anyway (and not writing changes to the database certainly should count as a failure).

This may be an edge case in the Zope3 database adapters transaction machinery - I suspect this is due to the lack of two phase commit in PostgreSQL 8.0. We need to update the PostgreSQL database adapter for Zope3 to take advantage of this feature once we have upgraded to PostgreSQL 8.1, ensuring that exceptions will never be raised when committing changes to the PostgreSQL database (which may occur after changes have been committed in the email subsystem).

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.