[Spacewalk-list] Spacewalk Proxy 1.7: SSL.py DEFAULT_TIMEOUT

[Chrzeszczyk, Jim]

Dear Spacewalk-list,

I'm running a distributed environment with a central Spacewalk server in the US and several proxies spread around the world. So far we've been on 1.5 and it was working fine with a couple of tweaks (mainly around adding PXE/TFTP on the proxies to make provisioning less painful)

Now I've installed a 1.7 Spacewalk (1.7s will eventually replace the 1.5 environment which was more of a proof of concept), I finished basic setup so I've kickstarted a Spacewalk Proxy from the "Master" Spacewalk and tried to provision a client from it - so I can continue testing kickstart profiles in my environment in Australia, as opposed to working remotely in the US.

This turned out to be a challenge - I couldn't make the installer download install.img in any way. I could see that the squid is downloading the whole file (I was tracking the connection in jnettop and it was going all the way up to 120-ish megs in around 10mins - my link between US and Australia isn't the fastest one in the World), but 2 minutes into the process, squid would log TCP_MISS/000 and the downloaded file would never make it to the cache.

It took me a bit of a research to figure out it's actually apache timing out, not squid (running a HTTP GET via telnet with no timeout made me realize that), and squid dumps the file cause the client doesn't want it anymore (000 code means more or less that the client interrupted connection as far as my investigation goes). I tried increasing timeout values in apache config (this failed miserably again) but when I started to google for rhn apache problems as opposed to rhn squid problems, I've found the solution very quickly:

http://comments.gmane.org/gmane.linux.redhat.rhn.satellite.user/1878

(...)

One can edit:

/usr/lib/python2.4/site-packages/rhn/SSL.py

and change the:

DEFAULT_TIMEOUT = 120

(...)

As it takes me 10 mins to download the 120MB file (and that's the biggest file I will ever need to download via the Proxy) I bumped that up to 1200 and the installation took off finally.

Now my questions:

1) why this wasn't a problem with 1.5 and became a problem with 1.7? (I  just doublechecked it on 1.5s - it's set to 120 as per the default setting which doesn't get in the way of downloading the same file over the same link just between a different pair of VMs)? Is this a bug or is this a side effect of some bugfix?

2) Is changing it to such a high value likely to cause me trouble in the future?

Best Regards,
Jim Chrzeszczyk