[Nottingham] Cups slow problem (printing long pause to start)

[John]

Ok, at home now so will try to reply in-line using Thunderbird ...

On 25/08/16 13:01, Martin via Nottingham wrote:
> Further thought inserted:
>
>
> On 25/08/16 11:48, Martin via Nottingham wrote:
>> On 25/08/16 10:48, John via Nottingham wrote:
>>> The original server hadn't been patched for donkeys, so at a loss why it
>>> started happening
>>
>> OK... Different aspect to consider for what might have changed...
>>
>>
>> Are you now printing new printouts that include some big data like
>> graphics or new fonts to be uploaded to the printer each time?
Nope. The output is a couple of A5 pages which in PCL is still a reasonable size.
>> That is, is the delay due to a lot of data being spooled up to the
>> printer first?
No, the delay is once cups tries to print to the printer
>> Server disk become full or badly fragmented?
>>
>> HDD disk errors causing a slowdown?
The server is virtual in our VMware farm connected to a venerable but excellent EMC San.
>> Are you connecting via WiFi and there is newly more WiFi activity?
Wash your mouth out :)
>> Sleep/power-saving mode been newly enabled on the printer?! :-P
The first print in the morning or after lunch is slow due to power saving / the laser printer spinning up,but subsequent ones should always 
be fast.
> Printer run out of local memory for new bigger prints or more print data?...
See the PCL comment above
>> And, can you see anywhere where there is high load? Or can you identify
>> that this is some sort of network timeout you're waiting for?
Yup. Running cups in debug mode shows cups hanging, hence the hypothesis it was snmp. Funnily enough, by the power of Grayskull (ok, vmware) 
I gave the server 8 cores,but cups had no better throughput.
>> Are your thin clients themselves waiting to pull down a config file or
>> to check the printer status as part of trying to print?
You're getting warm (not enough for a beer though(maybe a warm beer)). I ended up running a remote lpstat and that was hanging for the 10 
seconds for several thin terminals, but not for the centos / windows boxes.

>> Overly slow authentication delays??
>>
>> Slow DNS lookup?...
Slow DNS is always the first thing I check. Went through a period of not using nscd on the Linux servers, but it's running on them all now.  
My fault for running a *cough* Windows dns server.
>> Thin client memory swap/thrashing causing a pause?
It is a version of Linux (whatever was popular 15 years ago) and runs ssh & lpd. Nuff said.
>>
>>
>> Good sleuthing,
>>
>> and good luck?
Don't know if it was a power surge or just old age, but several of the thin terminals are exhibiting printing delays. Cups (why isn't lprng 
available anymore ?) gets hung up on these. Printing a document every 5 seconds should be a piece of piss, but enough devices hanging  
causes a backlog which affects them all. Again, 8 cores makes no difference. Running a suspect device in our office was still slow. Running 
a Centos / windows box was back to full speed.

So I've ordered 10 shiny new Core-I3 HPs to run Putty & Lpd in the short term & *cough* Windaz apps in future. Just need to pore through the 
cups logs & network events to target the misbehaving machines. BTW we use a utility called LanToplog which is very Ronseal and has been 
invaluable in checking the network structure.

Thanks to all for the suggestions. I just hope the management doesn't come to the conclusion that anything getting on in years should be 
replaced.
>> Cheers,
>> Martin
>>
>>
>> (OK, note the buckshot-at-barn approach for a beer ;-) )
The next round's on me. (Or is that the Milky bars ?)