[cups.bugs] Socket backend communication problem Kyocera IB-20IB-21 network cards
William Crowell
bill at crowellsystems.com
Thu Nov 9 12:50:09 PST 2006
Printer issue finally found:
If one prints files, say /etc/inittab, to the page printer using a RAW queue,
the first page will print. Subsequent pages die inside the printer. Some page
printers absolutely require CRLF sequences in the print when raw.
The error of a job being left in the queue was resolved by updating to cups
1.2.6. On cups 1.2.4, this error condition in the printer was manifested to
the spooler by leaving a file in the print queue.
I love my job. I love my job...
On Thursday 09 November 2006 14:55, Michael Sweet wrote:
> William Crowell wrote:
> > CUPS 1.2.6
> > GLIBC 2.4, 2.5
> > GCC 4.1.1
> > Kernel 2.6.18.2
> > (Gentoo Base)
> > All Kyocera printers using IB-20/IB-21 network adapters with socket
> > connection to 9100.
> >
> > First print job will process most of the time. Subsequent print jobs
> > produce no output. Something in the socket communication is causing the
> > Kyocera cards to hang. Power-cycling the printer will reset for the next
> > job.
>
> The socket backend does the following:
>
> 1. Create a socket
> 2. Connect the socket to the printer
> 3. Write the print data to the printer and read any backchannel
> data from the printer
> 4. Shutdown the outgoing side of the socket
> 5. Wait for the printer to close the connection, reading any
> backchannel data from the printer while it waits
> 6. Close the socket
>
> You can look at the code if you like, but it can't get any simpler
> than that. Either the Kyocera print servers are broken or the Linux
> TCP/IP stack is broken. My money is on the Kyocera print servers.
>
> You can try using the "waiteof=no" option to see if the Kyocera
> doesn't like the shutdown stuff (step 4 above):
>
> socket://ipaddress/?waiteof=no
>
> however, that would just indicate that the printer does not implement
> the TCP/IP spec properly and you need to get a firmware updated from
> Kyocera that fixes the bug.
>
> > *This is not a problem with running multiple copies of a single job.*
> >
> > /var/log/cups/error_log shows messages about "discarded events" from the
> > socket backend.
>
> That's the notification stuff, and they aren't coming from the socket
> backend, but from the scheduler which sends event notifications for
> changes to the job or printer, as appropriate...
--
William G. Crowell, VP & CTO
Crowell Systems
4235 South Stream Blvd Suite 100
Charlotte NC 28217
704.665.2000 fax 704.665.2180
More information about the cups-devel
mailing list