PaperCut Performance trouble shooting

[MrWu]

Hi all

 

Just working my way methodically to tackle a very intermittent PaperCut printing issue (MF 19.1)

 

Running on Server 2019, spec at end of post, brand new hardware and VM (1 x 500GB Gen2 VDMX running Win 2019 and PAperCut, no other roles) Copiers are Canon iR-ADV C5560 III, latest Generic PCL6 drivers, all Windows driver settings as recommended (TCP/IP Port, Winprint RAW, advanced printing features and Bi directional off)

Follow Me print queue all recommended settings (printer people checked this through)

Brand new HP 5412 Core and HP JG962A series edge switches, 10 Gb fibre to edge cabinets

 

Very occasionally (not at busy times strangly) for example, when a user sends a 30 copies of Duplex Black and White, it takes 10 seconds say per copy to come out?

 

Sods law I jump on every copier and it prints fine, and day to day (we only went live Monday though) its all quiet and happy, otherwise staff would have screamed if it happened all the time

 

I have Canon iR-ADV C6565 III high speed black and white copier I can test tomorrow (so far compliants are from 5560 III Colour model.)

 

Now, if a user released the print job on the copier, and the printer starts printing the first page and slow every page thereafter (This is 30 copies of the same document by the way), where is the bottle neck at? Would that be the Copier itself or PaperCut server trying to spool each duplex page to the Copier queue? Is the server HDD an bottle neck? (I did spec Tier storage with 3.2 TB SSD so should be ok?) Is there particular settings you folks do to tune the network card, virus scanning etc? The Hyper V Host connects to the Core on 10GBe

 

Any advice appreciated!

 

Server Spec:

 

Intel® Server SystemDual Intel® Xeon® 4114 Silver Processor 2.20GHz (13.75MB, 2S, 2 UPI, 85W)

10C/20TIntel S2600WFT Mother Board

256Gb Memory - (8 x 32GB 2400MHz DDR4 ECC Reg CL17 DIMM

Intel® Integrated RAID Module RMS3CC080

2 x Intel® SSD DC S4510 Series (240GB, 2.5in SATA 6Gb/s, 3D2, TLC) (O/S)

1 x HGST ULTRASTAR SN260 3200GB 3DWPD HH-HL PCIe

6 x Seagate 6TB 7200 RPM 3.5" 12GB SAS

Pioneer SATA Slim DVD-RW

Integrated Dual Intel® 10GbE Ethernet Controller

8 x 3.5" Hot Swap SATA/SAS Drive Bays

System Management Module 1300W AC Common Redundant Power

 

At the moment the Host hardly runs anything, One Domain controller, the PaperCut server and 2 test servers, thats it...

[Steve21]

Is this always with certain file types or any?

 

We had a very similar issue with PDFs in the past, only affecting Duplex printing over a certain amount of pages, and it was due to the photocopiers we had having a stupid (read -> Vendor own, rather than paying for proper ones) drivers on that really weren't happy with Duplex merging/flattening of PDFs

 

Anything directly from the photocopier (e.g. photocopies) worked fine as it didn't have to convert it, and any single sided pages printed also worked fine, it was just the duplex printing of PDFs etc that caused a hassle

 

And yet the higher end models of the photocopiers had no problems due to different (real) drivers etc

 

Steve

[MrWu]

Is this always with certain file types or any?

 

We had a very similar issue with PDFs in the past, only affecting Duplex printing over a certain amount of pages, and it was due to the photocopiers we had having a stupid (read -> Vendor own, rather than paying for proper ones) drivers on that really weren't happy with Duplex merging/flattening of PDFs

 

Anything directly from the photocopier (e.g. photocopies) worked fine as it didn't have to convert it, and any single sided pages printed also worked fine, it was just the duplex printing of PDFs etc that caused a hassle

 

And yet the higher end models of the photocopiers had no problems due to different (real) drivers etc

 

Steve

 

At the moment they are Word documents, duplex, black and white (very few staff reporting slow downs though .. but I’m sure it happens ) I changed print processor to winprint RAW and thought that solved it, until one teacher reported slow down this evening when printing duplex (26 copies of 2 page documents) ... I was thinking of using UFR 2 canon drivers instead of Canon PCL6 generic but I think Papercut recommends not using UFR 2

[Steve21]

This was the drivers on the actual MFP though, rather than print servers. Not even sure drivers being the correct phrase in this case (Page Description Languages etc), but the PS/PCL conversion software/module etc built into the MFP

 

In our case, it was linked to a paid for license/module that wasn't being provided that did the conversion, and instead a vendor-own version of the software doing emulation was in place which choked at anything more than a tuppence thrown at it

 

Steve

[MrWu]

This was the drivers on the actual MFP though, rather than print servers. Not even sure drivers being the correct phrase in this case (Page Description Languages etc), but the PS/PCL conversion software/module etc built into the MFP

 

In our case, it was linked to a paid for license/module that wasn't being provided that did the conversion, and instead a vendor-own version of the software doing emulation was in place which choked at anything more than a tuppence thrown at it

 

Steve

 

Thanks Steve will check this out too.

 

So in theory, if a single print job (despite say needing 30 copies) gets released from the Virtual Queue = Follow Me to the Physical Print queue of the copier and it starts printing, then has it technically finished spooling off the server in this case (The Copier driver is set to print once the last page finishes spooling, so in theory its passed it onto the copier once the first page comes out? So the following 29 copies of the document is handled by the copier (and not netork or the server?)

[Accura2k7]

Ours was very slow to start with on our Toshiba copiers, but was told its down to the driver not picking up the copier settings correctly... We have to make sure nothing it auto detect and the actual copier setting are set right... Goes full speed now!

[GaryPad]

If the MFD releases the first page and then drip feeds the rest, the problem won't be your print server or PaperCut - it's the MFD. MFDs typically buffer Mbs of print job so the vast majority of jobs should disappear straight from the Windows queue and get sucked into the machine. You can increase the storage on your MFD with an internal HDD, so there is only 1 instance of the job sent to the MFD which it caches and reproduces X times (depending on your collated copy count). Whilst I say this is the MFD it can also be the print job itself. So if the driver thinks the MFD has no storage for that 1 original the driver creates a concatenated number of copies, e.g. 10 copies produces a print job 10 times bigger than 1 copy (so storage on the MFD is critical). If that's the case then the job in Windows could be massive, take ages to spool, the transmission to the MFD will back up - and the whole process is painful. Check the PaperCut logs for the jobs that are slow and look at the size (attribs value). Look at the driver settings on the Find Me queue - do the Accessories match what the MFD has in terms of internal HDD? The driver might be creating concatenated copies instead of 1 original which the MFD multiplies. I would also avoid "Generic" drivers like the plague; look for monolithic drivers specific to that device, assuming that Canon do one for the iR-ADV C5560 III. Lowest common denominator drivers are always the least efficient.

[MrWu]

Thanks all

 

This has given me lots of pointers and support what I have been seeing.

 

So past 2 days I’m sure it’s the MFD

 

I saw 4 x print jobs sent to the Canon MFD and disappeared fairly quickly off both virtual and the actual device queue

 

I then logged into that copier via web browser and sure enough the 4 jobs were sitting there with the first job chugging away...

 

I asked the member of staff and she said everything from printing via PC to logging into the copier was super quick, until the JoB release bit..

 

I will do a bit of intensive testing with a test PC running with UFR2 drivers directly via TCP/IP to see what performance that brings ...