I am continually running into this issue. I have a Scientific Linux server with an HP printer attached to it. This machine is on a wireless network, as is a Vista Home Edition laptop (Toshiba Satelite). For months at a time, the printer works perfectly, without incident or loss of work, for jobs spooled from the laptop.
But every few months or so (sometimes weeks, it varies), the configuration stops working. Then I go through this drill of trying to figure out what the problem is, by testing for CUPS requests over wireshark, etc.
Usually, it turns out that I can ping the Linux server from the laptop, and I can reach the CUPS web interface from the laptop. But for some reason, the laptop cannot see the printer. I even try (several times usually) to remove and re-add the printer on the laptop.
I do not have Samba running on the server, and never have really (years ago I did run Samba on other Linux servers, but never needed it here). I do not wish to add Samba to the mix since printing from laptop worked previously without effort.
Wireshark does show ping and responses, and HTTP requests to the print web server, which I expect since those do work from the laptop. But I do not see CUPS requests or other packets for the printer.
I also checked that the firewall has the CUPS/IPP ports open, and made sure that I can print locally. Sometimes, I even reset the wireless router (that has cured it on one occasion!)
Why should this be so intermittent? I have not performed any upgrades here (other than the usual security updates for both Linux and the Windows laptop).
There are tons and tons of hits for this and very similar issues, but none of the solutions there work for me. And I have tried dozens of them (mostly ending up the same advice).
Sometimes rebooting Windows fixes it, but not this time. Resetting the wireless router did not fix it, and neither did removing and re-creating the device on Windows.
This is (intermittently) baffling and rather annoying.
Well, guess what? It turns out that the name of the queue on my CUPS server had changed! Once I reconfigured the laptop printer config to point at the new name, it works.
But, how and when did that happen? I didn't change it. I always call our one printer HPC4400, and that was the way it was configured on the laptop originally and it had worked that way for months.
I had no reason to change the name, and I certainly would not have chosen a long name like "Photosmart-C4400-series" which is the new name of the queue. Like I say if I had been forced to choose a name or rename it, I would have picked my much shorter name. This name looks like it was generated by the system somehow.
My print configuration had worked pretty much flawlessly for months. Why and how could this have changed without my intervention? There were no re-installs of CUPS on the server any time recently.
So, I did some MORE investigation, and I think I am seeing a big problem. Examining the boot log (last -x) for Friday night, I rebooted several times, in fairly rapid succession, while testing various combinations of USB hardware to determine why my webcam had caused a system hang, a system crash, and ultimately, knocking out some other USB devices, including my mouse, wireless, and ... usb printer!
Well, this starts to make some sense, or at least ties events together. So the next thing I did was examine the cups log (which, true enough, I probably should have done to start with). What I noticed is that my print queue called HPC4400 disappeared during the same time span that I had rebooted (on one of those several reboots per the boot log).
I can see, from the cups log, that there was a request for the print devices, which succeeded, during one iteration. A minute or so later, the boot log shows I had rebooted. This time, the same type of request for print devices comes back with an error! Later in the cups log, the print queue with the other name (not HPC4400) shows up.
So, it looks like my usb web cam knocked out the printer (along with my mouse and wireless), and that somehow caused cups to change the name of the print queue on its own. This still doesn't satisfy me, since under no conditions should cups be renaming print queues. I can only imagine what kind of havoc that might cause in a data center with dozens of printers and even more user client machines using cups.
I suppose this issue should now be redirected to the cups organization. If anyone here does happen to know of this issue, please feel free to post back here. thank you