[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [MirageOS-devel] Anyone else have problems with openmirage.org?



...and that was the bug, easily fixed now that you spotted it Nick :-)
Here's what's been happening with the live infrastructure:

- We've been running the bleeding edge version of MirageOS on openmirage.org
  with a stress test constantly running against it (a curl script in a loop)
  to help find memory leaks lurking in the TCP stack.  Several have been found
  and fixed, so this is all a good use of the live infrastructure.

- At the same time, the openmirage.org host was running in an old server in
  Cambridge University, and I migrated it over to Bytemark.  Cambridge hosting
  often went down due to some power instabilities here.

- Meanwhile, we're hard at work on preparing the Mirage 2.5 release which
  includes the Xen OCaml TLS stack.  This is deployed live on the new
  https://mirage.io site (for while we have a TLS certificate), and that is
  served from a fork of mirage-www that contains all the SSL serving logic.
  https://github.com/mirage/mirage-www/pull/317

- There was a mixup in the deployment scripts on the new server that led the
  openmirage.org to be compiled with the redirect logic from mirage.io, and
  so you're seeing all the https:// links.  I've now fixed this.

Andrew, if you're seeing the site hanging for some connections, this may be
due to some of the retransmit timer settings in the TCP stack due to you
coming in from a high latency high bandwidth connection.  I'm planning to
setup a permanent pcap trace on the live mirage.io site so that we can debug
and trace through these issues a lot more easily.  If you can get us a pcap
trace of a hang from one of your servers, that would be most helpful in
tracking this down.

cheers,
Anil

> On 21 May 2015, at 10:11, Nick Betteridge <buzzheavyyear@xxxxxxxxxxx> wrote:
> 
> Hi Amir,
> 
> Well, I was actually trying to read http://openmirage.org/blog/intro-tcpip as 
> I was initially trying to get some of the skeleton examples to work.
> 
> The above link didn't work, but does now. However, I've noticed that if I 
> read a http://openmirage.org and then click a link - your pinata article, for 
> example, then the actual link redirects to 
> https://openmirage.org/blog/announcing-bitcoin-pinata and my firefox tells me 
> 'unable to connect' 
> 
> Best
> Nick
> 
> Subject: Re: [MirageOS-devel] Anyone else have problems with openmirage.org?
> From: amc79@xxxxxxxxx
> Date: Thu, 21 May 2015 09:30:25 +0100
> CC: andrew.stuart@xxxxxxxxxxxxxxxxxx; mirageos-devel@xxxxxxxxxxxxxxxxxxxx
> To: buzzheavyyear@xxxxxxxxxxx
> 
> I noticed the same a couple of days ago. D'you remember which page(s) it was?
> 
> https://github.com/mirage/mirage-www/issues/321
> 
> Best wishes,
> Amir
> 
> --
> Sent via mobile
> 
> On 21 May 2015, at 08:32, Nick Betteridge <buzzheavyyear@xxxxxxxxxxx> wrote:
> 
> Over the last couple of days I got an 'out of memory' error on 
> openmirage.org/blog, yet open mirage  was fine.
> 
> _______________________________________________
> MirageOS-devel mailing list
> MirageOS-devel@xxxxxxxxxxxxxxxxxxxx
> http://lists.xenproject.org/cgi-bin/mailman/listinfo/mirageos-devel
> _______________________________________________
> MirageOS-devel mailing list
> MirageOS-devel@xxxxxxxxxxxxxxxxxxxx
> http://lists.xenproject.org/cgi-bin/mailman/listinfo/mirageos-devel


_______________________________________________
MirageOS-devel mailing list
MirageOS-devel@xxxxxxxxxxxxxxxxxxxx
http://lists.xenproject.org/cgi-bin/mailman/listinfo/mirageos-devel


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.