Last zope error:

Prints, but never reads 'printed' variable.
Doesn't print, but reads 'printed' variable.

am I the only one who sees some little inconsistency in this double-error?

Posted by mattia as zope at 3:50 PM CEST

No Comments »

Got into the big adventure. Just a fast advice: if you start getting ClientDisconnected errors, stuff about “database conflict error” or “serial this txn started with None…”, then simply erase both *.zec files on zope clients and all Data.fs.* (not Data.fs itself, please!!!) on the zeo instance. You’ll have better luck afterwards.

Posted by mattia as cluster, database, zope at 3:20 PM CEST

No Comments »

It’s a problem when you write diagrams like this. It’s a bigger problem is you really understand them.

Posted by mattia as apache, cluster, diagram, sketch, squid, zope at 12:57 AM CEST

No Comments »

I just learned this:

Of course. Locally generated packets don’t go into the PREROUTING chain.

(from this post)

My thought about this is: you should all burn in hell. You, who decided this rule, in the deepest, fiery of hells, eaten by the cruellest devils, half of your body in black ice and half in an even blacker flame.

If local generated packets don’t go through prerouting, DNAT to localhost is always *half* performed (when the packet enters is DNATted in prerouting, but when exits it no longer passes through prerouting and DNAT can do nothing).

So. WTF someone decided local generated packets don’t go through PREROUTING?!?!

Posted by mattia as dnat, iptables at 5:25 PM CEST

2 Comments »