Error -3 while decompressing data - SOLVED

[Liam Routt]

I'm finally able to, I think, resolve this issue.

We've done a bunch more tests, and today began pulling out hardware 
components. Low and behold our problems disappear when a particular 2GB 
stick is removed from the machine, and return whenever that stick is 
present. So it would seem to be an unlikely hardware fault.

We've tested the memory stick with MemTest and it found nothing 
untoward, but the real-world test results are 100%. We've run four (or 
more) successful tests since removing the memory, after never having a 
successful run before. We've returned the memory and had a failure again.

We also grabbed the new bzr 2.3b5 and have run that fine (without the 
memory), so that's what we're going to use at the moment.

In the end I'm glad I didn't go ahead and report the fault on launchpad. 
I was never entirely comfortable with what we were actually reporting, 
and I'm glad not to have muddied the water.

Thanks for all the help we got from folks here. Glad to be able to keep 
using this great system!

Take care,

Liam Routt
Media Saints

On 13/01/2011 6:44 PM, Liam Routt wrote:
> We've run some more tests, and I feel we are getting much closer to
> working out where the problems lie for us. We're going to run a few more
> tests over the next day before we make the report, as it really feels as
> though we are almost there to working out where the nut lies, which will
> make a better report.
>
> In summary, though, we found out:
>
> * The bazaar server and communication with it is not the issue in itself
> - the problem happens using the filesystem directly (I was wrong on my
> recollection of the tests we'd done).
>
> * It does not fail when running on an roughly equivalent machine with
> 32-bit kernel (same distro, OpenSUSE).
>
> * It does not fail when using the 32-bit install over the existing
> 64-bit kernel on the original machine (ie, copied the install from the
> second machine to the first machine, and ran chroot-ed, and the test did
> not fail).
>
>
> This suggests that the kernel isn't the issue, but that the 32-bit
> version does work, where the 64-bit one doesn't work.
>
> We're going to do some more tests to pinpoint what factor is the
> controlling one here. For starters we're going to run the tests on
> another 64-bit install (also OpenSUSE) to make sure it isn't some local
> install problem with this machine somehow. And we'll take it from there.
>
> Take care,
>
> Liam Routt
> Media Saints
>