download problem experienced by Zenoss community member

[gstreiff]

One of our Zenoss community members is unable to download certain document files we've recently published.

His experience is described here:

http://community.zenoss.org/message/66473#66473

Community member has tried exhaustively to pull the documentation down via multiple systems and browsers.

Interestingly, he can download content under:

http://sourceforge.net/projects/zenoss/files/zenoss-beta/builds

but no joy with the *.pdf files directly in:

http://sourceforge.net/projects/zenoss/files/zenoss-beta/

An example of a file he cannot download is:

http://sourceforge.net/projects/zenoss/files/zenoss-beta/Zenoss_Core_Installation_01-052012-4.2-v04.pdf/download

He is getting 206 errors.

Something is unique here as other folks area able to download these pdf's so suspect something amiss with the mirrors.

I'm going to re-touch the content from my shell account in an effort to trigger the mirroring again.

Would you please investigate? Thanks for your help.

[ctsai]

Greetings,

Okay, so I've read over that thread a few times now and here are my thoughts.

1) As with you and several other users, I can't reproduce this issue.

2) Jo could try using Firefox in Safe mode, and/or Chrome in incognito mode. This would be an attempt to see if it's a browser plugin or something like that causing some sort of interference.

3) I would also like to see if this issue is reproducible for him with *any* PDF file. I uploaded a test pdf here he could test: http://sourceforge.net/projects/strawhat/files/sf-frontpage.pdf/download

4) A screencast of the exact behavior he's seeing, or a error log of some sort might be useful.

5) A "definitive" list of mirrors tested might be helpful as well. I see in his last post that he's still having issues with Softlayer, though I just tested the one you linked to above from Softlayer and didn't have any issues. With that said, it's possible that there's an issue with configuration on specific mirrors.

In the end though, one thing I believe is obvious is true is that this is either a SourceForge issue, or an issue with the user's particular setup(s). This is *not* a "Zenoss" issue. That is to say, it's only your issue as far as one of your users is affected, but it's not necessarily yours to fix. As such, I would recommend that Jo communicate with us directly (probably via this ticket) to continue troubleshooting.

Let me know if this helps.

Regards,
Chris Tsai, SourceForge.net Support

[gstreiff]

Replying to ctsai:

Let me know if this helps.

Regards,
Chris Tsai, SourceForge.net Support

Thanks for your helpful reply, Chris. I'll pass this along to Jo via our community forum so he can follow-up with you as he desires.

Internal testing yesterday at Zenoss surfaced issues with at least these three mirrors:

CityLan (Moscow, Russian Federation, RU)
Softlayer (Dallas, TX, US)
garr.it (Ancona, Italy, IT)

Additionally, Jo had issues with 6 mirrors and specifically identified:

Ireland/HEAnet
Softlayer

At this point, Ireland/HEAnet was working for him as of yesterday.

Glenn

[jrhett]

If you read the thread, the problem was reproducable on 4 different browsers, 3 different operating systems and computers, etc etc etc. And confirmed using wireshark looking at the network traffic layer. That's not a browser problem. Most of the browsers I tested with have NO plugins, since those are the ones I test websites with for $dayjob.

[ctsai]

jrhett,

Okay, so after further testing, I was able to reproduce this with the following:

Mac OSX Lion, Chrome Version 19.0.1084.52 (in incognito mode), using softlayer with this file:
https://sourceforge.net/projects/zenoss/files/zenoss-beta/Zenoss_Core_Installation_01-052012-4.2-v04.pdf/download?use_mirror=softlayer

When I tried that exact link in FF or Safari, it worked fine. When I tried using a different mirror in Chrome, it worked fine as well. In any case though, I'm escalating this to our operations team for further review.

Regards,
Chris Tsai, SourceForge.net Support

[bprew]

Issue appears to be overloading at the mirror itself. We've improved our mirror testing so it will take a given mirror out of rotation faster when it becomes overloaded, but as it is a transient load issue, it may occur occasionally.

Thanks

[gstreiff]

Thank you for investigating.

Glenn