I don't know if the developers are still working on the SQUID source or 
anything.. But here is a NIFYT idea?!?! well at least its an idea towards a 
single (main) process squidguard.

Build a client/server structure to the redirector, using TCP 
ports.  Basically have the squid-redirectors communicate all data to a 
single process running completely self-sufficient.

Then if squid SIGHUP squidguard is immune to the effect, only the 
children-client processes restart.

The server can then be SIGHUP independently without any ill effect.

To handle the time that the server-squidguard process may be out of 
commission while it reloads and the clients are still sending requests, 
have built into them a default "RELOADING.. error page, long the lines of a 
err 500, or so."

This way the clients sending new requests will get the error, but clients 
in the process of transferring data will be unaffected.

I know I am no programmer with the background to write this, that's why I 
decided to post the idea to the mailing list instead.

But this may improve the ability of squidguard's system since it now can be 
made internally multi-threaded, as well as keep a single db in memory, etc. 
but not kill the clients.

Ideas, comments? Feel free to share!!


Reply via email to