Apache OpenOffice (AOO) Bugzilla – Issue 5049
User handling on website broken
Last modified: 2003-12-06 14:52:32 UTC
1. Administer users: no matching user found. 2. If I leave my session open, I do not login again (e.g. for entering issue) but I changed my user id. I guess both issues are related.
This also means, that we are not able to change roles of a user. P2 -> P1
reassigning to support
The issue with the initially blank Administer usres screen has been filed with an engineer for investigation. The users are available if a search is conducted on this screen for a particular keyword or user name, which should allow you to modify roles. I am not sure I understand your second point, could you please describe further? Thank you
I logged in as mh, worked for a little while, left the application for some hours idle, then clicked on "my Issues" and saw the bugs of someone else. clicked on "edit profile" and saw the profile of somebody else.
I'm still looking for a workaround to assign cmc@openoffice.org the project lead role in the sw project.
ok, the work around is: - make the project owner role requestable - send mail to person that he/she rquest role - approve role - reset requestable role again not nice but works.
Thank you for the explanation, and I am glad you were able to workaround the problem of adding a user to a project. An engineer is looking into the larger issue and we will hopefully have a solution soon. I am still looking to reproduce the second issue you raise of being presented with another user's issues and profile without changing your login information.
I have not yet been able to reproduce the error of seeing another user's information after being idle on the site for a period of time. I am always requested to log in again. The Administer Users issue is with an engineer .
I'm also not yet been able to reproduce the error of seeing another user's information after being idle on the site for a period of time anymore.
Interresting. I'll try a bit more and see if anything comes up. Continuing to investigate the user handling issue next week.
fwiw, I've had some people register, login, and then find that SC has interpreted their logins as somebody else's.... quite disturbing. cheers louis
Louis - if you can see a common thread for that behavior please let me know. Further work will be done this week on the staging server pertaining to the main administer users issue.
I have another guy with problems here. He can't submit issues. His login is pde (Peter Eckersley). I can find him as a registered user, when I use the user administration. But I can't find him as a user in IssueZilla, which is certainly the reason why he can't submit bugs. Just wanted to mention this, in case it's related to this bug.
That is odd. Do you know when this user was created (recently/long ago)?
my guess and it is only a guess is that it may be necessary to run a script to update all the user databases. We can do this perhaps when we migrate to Solaris. louis
Changing whiteboard to reflect issue council marker. It is possible the user was "lost" because of optimizations made to the issue tracker rebuild process to deal with the large number of users on the site. Triggering a full rebuild may resolve this issue but would impact site performace while it runs. Durring the migration all databases should be updated.
The launch of the staging server has been delayed by two days and is currently planned for Tuesday of next week. We appologize for the delay.
Myself and Louis experienced blips in identity recently. Issue pcn9876 has been entered for investigtion of this phenomenon. There are no further developments to report on the main user handling issue.
The staging server has been built and populated with data. The domain change work (ref to openoffice.org -> solstage3) is underway now. The process of copying data and making the changes has taken longer than originally expected, and the server should be available tomorrow rather than today as planned.
Periodic sandbox restarts have been initiated as an interrim solution for the problem with user identity confusion. A permanent fix is being investigated by an engineer.
We have not experienced this issue so far on the staging server. We will continue to watch for this behavior.
Investigation on the production server has uncovered evidence that the problem has to do with the number of results being returned from the database. We continue to look into this and the relationship to the version on the staging server. .
Actually mistaken in my last statement. The problem with retrieving large amounts of data from the database lies a few layers below any SC code. The difference in platform is probably what is being reflected.
The user administration issue has been addressed by resetting arguments to the JVM. A listing of users is now displayed for administration. The user confusion issue is still being investigated.
Debugging scripts have been installed on the server to help diagnose the user confusion issue. Developers will be looking into what they have gathered on Monday.
Nothing significant has yet been found in the files generated by the debugging scripts. We are continuing to look into this.
No incidence of the ghosting behavior has been caught by the script to this point. As soon as an incident is captured debugging will proceed.
This behavior has still not been captured.
The watched error never appears. The user confusion behavior has still not been captured.
We have not captured this behavior in a week without sandbox restarts, and suspect that it was resolved by the change made to the JVM allowing administration of users. I am closing the issue at this time. If the behavior resurfaces it can be reopened.
removing from top5 and closing