maemo.org Bugzilla – Bug 2620
installing packages fails repeatedly with file "corrupt" error message
Last modified: 2009-05-19 11:56:03 UTC
You need to log in before you can comment on or make changes to this bug.
STEPS TO REPRODUCE THE PROBLEM: browse to maemo site and try to install canola2 by following the steps. Same thing randomly happens with some other apps. EXPECTED OUTCOME: packages/apps install properly ACTUAL OUTCOME: app manager says file is "corrupt" REPRODUCIBILITY: always (always/sometimes/once) EXTRA SOFTWARE INSTALLED: none OTHER COMMENTS: cannot really use N800/OS2008 due to inability to install packages. I reflashed the N800 but no improvement.
repository.maemo.org seems to be mostly yielding, in place of .deb packages, error pages with the wrong 200 OK http response, so application manager successfully gets garbage and complains correctly. testcase: http://repository.maemo.org/extras/pool/chinook/free/p/python-runtime/python2.5-runtime_c1.0-2_all.deb http://lists.maemo.org/pipermail//maemo-developers/2007-December/013458.html says: The correct behavior would be to send a "503 Service Unavailable" or other error message inside header, to inform correctly the apt utility.
extras-devel is also hosed: http://repository.maemo.org/extras-devel/dists/chinook/non-free/binary-armel/Packages.gz Ferenc seems to be around, hopefully he can take a look at this soon (Ferenc, hope you don't mind being added on cc!)
Just my objective 2c... it would seem to be a hosed http handler on the server. Test case : Browse http://repository.maemo.org/extras/pool/chinook/free/q/quiver/ Click on the quiver_0.1.18-1_armel.deb (it will download everytime) Click on the quiver_0.1.19-1_armel.deb (it will error everytime) So some logic wrapped around http requests is interfering.
*** Bug 2636 has been marked as a duplicate of this bug. ***
*** Bug 2635 has been marked as a duplicate of this bug. ***
Note that the content answer depends on the effective akamai IP for repository.maemo.org.
This is impacting all sorts of applications both new and updates. NEEDS COMMENT AND RESOLUTION!
It actually seems to have gotten better yesterday. I was finally able to install canola2 and UKMP and others...
As a temporary workaround, I've found that changing my sources.list to point to stage.maemo.org instead of repository.maemo.org has allowed me to keep working, for some reason.
(In reply to comment #9) > As a temporary workaround, I've found that changing my sources.list to point to > stage.maemo.org instead of repository.maemo.org has allowed me to keep working, > for some reason. I can confirm this. After changing the repository to stage.maemo.org I could install Python2.5-runtime, which earlier failed every time and was blocking the install of several other programs.
(In reply to comment #10) > (In reply to comment #9) > > As a temporary workaround, I've found that changing my sources.list to point to > > stage.maemo.org instead of repository.maemo.org has allowed me to keep working, > > for some reason. > > I can confirm this. After changing the repository to stage.maemo.org I could > install Python2.5-runtime, which earlier failed every time and was blocking the > install of several other programs. > this solution works for me too
*** Bug 2610 has been marked as a duplicate of this bug. ***
Created an attachment (id=685) [details] Temporary Workaround for installing the SDK This is a temporary hack that works around the repository problem when installing the SDK. At present, if someone tries to install the SDK it is unable to get the source/Sources.gz file because the server reports "Service unavailable". However, this file is available if repostory.maemo.org is used instead of repository.maemo.org This patch adds a line to the shell script in the maemo SDK binary. This line modifies the sources.list file and replaces 'repository' with 'repostory' Your mileage may vary. This is my first time installing the SDK so there may be complications I'm not aware of.
I'm rather amused at this 'repostory' workaround, it originated on the ITT forums, and it has been repeated verbatim in several places, with no one bothering to understand _why_ it worked: repository.maemo.org points to the Akamai cache, and *.maemo.org points to the actual server, which is the same reason that stage.maemo.org works, which was pointed out on the mailing lists (I think). repository.maemo.org is "bad", and everything else is "good". People have been repeating 'repostory' like some kind of magic incantation. I realize there are many people who have not said anything who understand fully why this works, but it touched on a pet peeve of mine to see this endlessly (and IMHO mindlessly) repeated verbatim, which suggests the number of people who have repeated it verbatim have no understanding of the underlying mechanism. Frankly, this whole mess is a good reason why Nokia should dump Akamai for this kind of thing and do as their upstream Debian does and make use of the open-source mirror resources available. Akamai is good for streaming music videos and high-demand websites and whatnot to IE and Firefox, but for APT, which expects a bit more standards-compliant and robust HTTP behavior, it's just a pain in the ass.
*** Bug 2681 has been marked as a duplicate of this bug. ***
I was able to pull a random 203 packages from extras/pool/chinook/free just fine today (looks like I got 217.212.252.80) so this seems to wfm now from here at least?
Yes, I can confirm too. It's working fine now.