Wishlist: Better handling of backend error

[Michael Sweet]

Hatto von Hatzfeld wrote:
> Michael Sweet wrote:
> 
>> Hatto von Hatzfeld wrote:
>>> When a "backend did not respond within 30 seconds" the cups server is not
>>> reachable, even other printers are unusable, and it is very difficult to
>>> figure out what is wrong (cups does not tell which backend has this
>>> problem). I suggest that this should be improved.
>>> ...
>> This message is only displayed on cupsd startup - once the server
>> has started, normal processing can continue.
>>
>> That said, CUPS 1.2 does away with the probe on startup, and instead
>> only probes when a CUPS-Get-Devices request is done.  In addition,
>> the probe starts a separate cups-deviced process which does the
>> actual probe and allows the cupsd process to continue accepting
>> and processing requests while it figures out the available devices.
> 
> Then why the following?
> 
> ~> lpq
> lpq: Unable to contact server!
> ~> lynx http://localhost:631/
> (error: no connection to remote host)
> ~> pstree | grep cups
>      ├─cupsd───cupsd───serial

cupsd is just starting up - it is not in the normal operation mode
yet.  In this case, it is busy probing serial ports.

Like I said, this doesn't happen during normal operations, and in
CUPS 1.2 we actually run the device probe in a separate process to
avoid this issue.

-- 
______________________________________________________________________
Michael Sweet, Easy Software Products           mike at easysw dot com
Internet Printing and Document Software          http://www.easysw.com