Execute command (doc to pdf converting)

Rossini · Post by **Rossini** » Tue Apr 03, 2012 3:03 pm

Hi!

We´re using libre office to convert doc-files to pdf.

Element is "Execute command" and config:

Arguments: -headless -convert-to pdf:writer_pdf_Export -outdir "%2" "%1"

Output: File at path

Copy inoput job: Yes

Output extension: PDF

My questions are:

1. Is there a way to speed up the converting other than to edit the setting "scan intermediate folders"? We use recommended from Enfocus 300 sec and that will be to slow in this flow. If I lower this setting it will affect performance for other flows.

2. Also having trouble with result that leaves a jobfolder instead of just the pdf-file. I then have to use job dismantler that also uses settings from "scan intermediate folder".

The converting takes about 10 minutes in this case and I have to do it faster..anyone having some ideas to solve my problems?

Best regards

Henrik

dkelly · Post by **dkelly** » Tue Apr 03, 2012 7:00 pm

Is the conversion process taking too long or Switch taking to long to start processing a new job?

Rossini · Post by **Rossini** » Tue Apr 03, 2012 9:10 pm

It is SWITCH taking long time. I tried to set value to scan intermediate folders to 30 seconds.

It was of course faster but real time from scan folder before the element "execute command" was about 90 seconds and approx same time in the folder before job dismantler.

dkelly · Post by **dkelly** » Thu Apr 05, 2012 12:15 am

Something doesn't sound right. Could you send me a copy of your flow and I'll investigate?

Dwight Kelly

Apago, Inc.

dkelly@apago.com

Rossini · Post by **Rossini** » Thu Apr 05, 2012 8:12 am

Something doesn't sound right. Could you send me a copy of your flow and I'll investigate?

Dwight Kelly

Thank you, I´ll send a copy at once!

Henrik

dkelly · Post by **dkelly** » Tue Apr 10, 2012 6:19 pm

The maximum time to process a job is the sum of:

Submit hierarchy

1) Submit Hierarchy Scan every - 15 secs

2) Preferences -> Process files only after - 5 secs

pdf_in folder

1) Preferences -> Scan Intermediate folders every - 30 secs

2) Preferences -> Process files only after - 5 sec

So it could take up to 55 secs (15+5+30+5) before Libre Office is called to convert the .doc to .pdf

Here are my suggestions:

a) Reduce the "Scan Intermediate folders every" setting from 30 to 5 secs.

This will increase the load on your server because will be looking for new files in each intermediate folder much more frequently. However, files will move through the flow much faster. I have suggested to Enfocus that they add support for filesystem notifications for local directories. This would allow the OS to notify Switch will a file is created/modified/deleted instead of Switch constantly polling the directories and files for changes.

b) If the files are all on a local server, you could reduce the "Process files only after" setting to 1 sec.

Files aren't processed until their file size stops changing, eg. file is being written/copied. If all of the files are local, this will happen very fast and there's no need to wait an additional 5 secs to process the file.

c) write a custom Script element to replace the Execute Command element. This would allow multiple files to be converted at the same time, assuming there's sufficient memory/disc capacity on server

---

Dwight Kelly

Apago, Inc.

dkelly@apago.com

Rossini · Post by **Rossini** » Fri Apr 13, 2012 9:38 am

I replied to Dwight:

Our hotfolders are network volumes and Enfocus recommended us to set ”scan intermediate folders” to 5 minutes for maximum performance.

I tried with 30 sec and it gets faster then. But I´m afraid that could affect other flows?

“Process file only after” is set to 10 sec and “scan input folders” to 15 sec.

We tried some different configurations and decided this is ok.

Answer from Dwight:

It will increase overall load on both Switch and file servers. But not enough to cause any concern with other flows. Setting it to 5 minutes is really going to slow down the input of new jobs.

My conclusion:

I set scan intermediate folders to 30 seconds to start with and then we reduce the flowtime with about 10 minutes compared with yesterday. If thats not going to work we have to develop a custom script element allowing multiple conversations to run at once.