nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

DD multiple deployments

jkm

Hi all,

We deploy content from dcrs to database tables and some times number of records to push are more than 2000 while doing production deployments. This increases our production deployment time enormously.

Currently we are invoking datadeploy for each dcr individually means invocation of dd 2000 times and pushing the content from dcr to database. This increases the deployment time a lot. Is this possible to feed all the dcrs in single deployment rather than invoking dd multiple times and push the values of each dcr as different records in the table?

Your inputs are appreciated.

TS6.5+sp2
OD6.0.1+sp1 Solaris 9

Find more posts tagged with

Comments

nipper

Out of the box no. But you can use a scheduled deployment and run it every hour.

You can also group the DCRs into a single job and then it *should* be in a single deployment. Of course I did not write your WF so I do not know if that is indeed how it was coded.

jkm

Thanks ande ...this can not be scheduled because while doing production deployments we need deploy complete content in very tight window like 30-40 mins which is not enough to deploy 2000 pages using individual dd calls

I have already gouped all the dcrs in a single job and we loop for each file and then invoke the dd for each dcr ....invoking dd for each call cause es the pain...
This is coming as strong contention while going for production deployments in restricted timelines

nipper

Then your workflow is written poorly.

The deployment process should group like DCR, write their names to a temp file, after parse all files in a job, it should run open deploy for HTML files, open deploy for JSP (if you have an app server) and datadeploy for each different table affected.

One at a time is a horrible design.

Andy

jkm

sorry Ande to claim this a horrible design and poorly written workflow so quickly. I dont want to offend anybody here but to claim something without appropriate information ...may not be able accept

would you let me know how do you group the dcr files. If you need to push the content from multiple dcrs then u need to pick each file individually ..if there is any other way to pick all the dcr files in one go then it was my original question which u have said "No"....

nipper

It is a poor design. The external task that runs the deployments should have a loop that reads all attached files to the job and places them into the appropriate buckets (DCR type, deployment server, etc). You can filter DCRs by the category/type of the DCR. Put those names in 1 file and run 1 file list deployment with that file.

If you run more than 1 deployment for a specific job to the same location (docroot, server, db table, etc) that should change. If you submit 2000 files it breaks. Does that sound like a good design ?

Adam Stoller

It is a poor design. The external task that runs the deployments should have a loop that reads all attached files to the job and places them into the appropriate buckets (DCR type, deployment server, etc). You can filter DCRs by the category/type of the DCR. Put those names in 1 file and run 1 file list deployment with that file.

If you run more than 1 deployment for a specific job to the same location (docroot, server, db table, etc) that should change. If you submit 2000 files it breaks. Does that sound like a good design ?

Hmm - got up on the wrong side of the bed this morning Andy - or just tired of being called Ande ? (or maybe sorry you don't have a Mac?) :-)

How's this for a re-phrasing:

JKM - read the Database Deployment for OpenDeploy Administration Guide - search for all occurrences of the term "filelist" - then fix your config and your externaltask script to take advantage of it.

nipper

Hmm - got up on the wrong side of the bed this morning Andy - or just tired of being called Ande ? (or maybe sorry you don't have a Mac?) :-)

So you think that having 2000 datadeploy calls for a single job is a good thing ?

Adam Stoller

So you think that having 2000 datadeploy calls for a single job is a good thing ?

I never said that - I simply re-phrased what you said in terms of pointing towards a better solution leaving out the critique (I usually save that for when actual code is posted as opposed to just described ;-)

jkm

thanks guys ...currently in our ddconfig file we are using source path like this

[html]

[/html]

and if I use "filelist" parameter instead of "name" then it should do the trick.
I have not tested this still but looks like a way forward. To create a filelist of all the required dcrs will be a small job...

jkm

Hi All,

I made couple of fixes to improve the huge content deployments like

1. filelist parameter instead of path to reduce the number of dd calls
2. set the real-update parameter
3. row-map-cache-file parameter

This has improved the performance significantly but still I think there are chances of improvement. Another area I can think of logLevels

Any other suggestions are welcome to improve the dd deployment please.