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; } } }

DAS Performance

fiquebem

We have TS 6.5.0 and OD 6.0.1 on Solaris 9. We want DAS to deploy DCRs (only) to Oracle 9i. I'm experimenting with the example internet/careers DCT. I enabled the deployment of the replicants in it too (up to 6 "Responsibilities" and up to 6 "Skills").

The problem is that it is taking 10 minutes to run either the deltaupdate or the submitlist deployment (via DAS). Below is the top section of the deltaupdate configuration.

Incidentally, I'm seeing a whole bunch of entries such as the following in odbase log: DD:EventHandler class com.interwoven.dd100.syncdb.TDestroy. What does it mean?

DD:Generate DirectoryContents start:Fri Oct 21 13:46:24 EDT 2005
DD:EventHandler class com.interwoven.dd100.syncdb.TDestroy
DD:EventHandler class com.interwoven.dd100.syncdb.TDestroy
DD:EventHandler class com.interwoven.dd100.syncdb.TDestroy
DD:EventHandler class com.interwoven.dd100.syncdb.TDestroy
DD:EventHandler class com.interwoven.dd100.syncdb.TDestroy
DD:EventHandler class com.interwoven.dd100.syncdb.TDestroy
DD:EventHandler class com.interwoven.dd100.syncdb.TDestroy
DD:EventHandler class com.interwoven.dd100.syncdb.TDestroy
DD:EventHandler class com.interwoven.dd100.syncdb.TDestroy
DD:EventHandler class com.interwoven.dd100.syncdb.TDestroy
DD:EventHandler class com.interwoven.dd100.syncdb.TDestroy
DD:EventHandler class com.interwoven.dd100.syncdb.TDestroy
DD:Generate DirectoryContents end :Fri Oct 21 13:49:24 EDT 2005

Why does it take 3 minutes to "Generate DirectoryContents"?

Then the log shows the following:

DD:Generate TDCRTupleProducer start:Fri Oct 21 13:49:24 EDT 2005
DD:Generate TDCRTupleProducer end :Fri Oct 21 13:54:25 EDT 2005

Five minutes to "Generate TDCRTupleProducer"? Not nice. What to do?

Incidentally, we have the following in iw.cfg:

log_createfse=on
log_renamefse=on
log_modifyfse=on
log_destroyfse=on
log_syncdestroy=yes
log_syncrevert=yes
log_renamefse=yes
log_setea=no
log_deleteea=no

How to configure TS 6.5 and/or OD 6.0.1 to optimize DAS deployments? I suppose we can try using a "row-map-cache-file," but I doubt that that will make a big difference.

Thoughts?

Thanks,

S.

-----------------------------

<deployment name="deltaupdate">
<source>

<teamsite-templating-records
options = "full,wide"
area = "$myarea"
base-area = "$mybasearea"
>
<path filelist = "$myfilelist"
visit-directory = "deep"
delete-after-use= "yes" />
</teamsite-templating-records>
</source>
<destinations>

<database use ="oracle-db"
update-type = "delta"
state-field = "state"
>

Find more posts tagged with

Comments

fiquebem

I ran the command DAS is running behind the scenes, i.e.,

iwodcmd start ...careers_dd.cfg -k iwdd=deltaupdate ...

and it took about 9 minutes.

However, the equivalent deployment invoked via iwodstart, i.e.,

iwodstart ...careers_dd.cfg -k iwdd=deltaupdate ...

ran almost immediately. Ironically, iwodcmd is supposed to be more efficient. What could be the problem?

Can I make DAS use iwodstart instead of iwodcmd? If so, how?

Thanks,

S.

fiquebem

Actually, today I ran standalone deployments using both iwodstart and iwodcmd. They took the same time: 10 minutes. (Yesterday iwodstart was failing because of a missing parameter, but it seemed to be failing faster than iwodcmd.)

iwodcmd deploys those DCR files to the local file-system really fast. It's the database deployment (the "Generate DirectoryContents" [whatever that is] and the "Generate TDCRTupleProducer" tasks) that is taking curiously long.

Incidentally, we're running OD 6.0.1 and TS 6.5.0 + Search, but without any patches or Service Packs. (Unfortunately, we can't upgrade at this time.) We're noting that the Search server is hogging up the CPU. Wonder if that's impacting the database deployments (the file deployments seem to be unaffected).

Thanks for any suggestions,

S.

fiquebem

We made progress by changing the "source" element from "teamsite-templating-records" to "xml-source":

<source>

<xml-source
options = "full,wide"
area-type = "os-filesystem"
area = "$myarea"
xml-type = "interwoven"
>
<path filelist = "$myfilelist"
visit-directory = "deep"
delete-after-use= "yes" />
</xml-source>
</source>

Recall, we are trying to deploy DCRs to Oracle, via both DAS and standalone deployments. Now the standalone (manually initiated) deployments complete in about 1 second, but the DAS deployments still take too long. Once the DD Thread starts the DAS deployment, it takes only about 1 second. But there is a noticeable delay in the event handling/messaging somehow. What can we do to speed the DAS deployments? (We already have the TS Events Subsystem enabled.) I looked into daemon.cfg to see if there were any parameters I could change to get OD/DD to respond to TS events more nimbly. No such luck.

How long does it normally take for a simple DCR Save to be reflected in a database such as Oracle? How about a DCR Submit?

Thanks,

S.

fiquebem

Rebooting seems to have done the trick -- Now DAS deployments are reasonably fast. What's more, now the "teamsite-templating-records" source is working just as well as "xml-source."

However, sometimes there's a delay between an action (e.g., DCR Save) and the corresponding DAS deployment. Perhaps it is due to lack of caching sometimes.

But even with command triggers, sometimes there's a delay between an action and iwevents.log showing the corresponding event. Wonder why.

At the moment we've decided not to use DAS. That's because out TS Events Subsystem seems to go down every now and then. Besides, with the Events Subsystem and DAS, we're noticing duplicate event notification/handling.

S.

Migrateduser

Thank you for the updates on your progress.

Regarding the duplicate event notification. Can you elaborate on that or attach the log that shows this? I'd like to get a better picture on what's going on. Thanks again!

Mariam