[cups.general] When should a network backend retry?

[Michael Sweet]

On Oct 30, 2009, at 2:23 AM, Johannes Meixner wrote:
> ...
> Are failed lookups or routing errors really fatal errors
> in any case nowadays?

A failed regular hostname lookup (printer.example.com) means that your  
dedicated DNS services are either not responding or the hostname has  
gone away. There may be some changes for us to make in this arena,  
e.g. failure to contact the DNS server should be a retry instead of  
"stop this printer", but if the hostname is no longer defined there is  
no possible recovery without the admin fixing the DNS or printer  
configuration (which require the backend process to be re-run to pick  
up the changes...)  In any case, we'll need a bug on cups.org to track  
any change to the CUPS code... :)

If you use a dynamic host or service name (for CUPS that means  
a .local hostname or a service name that we resolve at print time), we  
retry the lookup until it succeeds since those names are intended to  
be used in dynamically changing networks.

Failed routing is tricky; we already retry for EHOSTDOWN and  
EHOSTUNREACH, but not for ENETDOWN or ENETUNREACH...  I've not seen  
any bugs on cups.org on this being a problem, though...

___________________________________________________
Michael Sweet, Senior Printing System Engineer



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cups.org/pipermail/cups/attachments/20091030/4d72bd3e/attachment-0001.html>