This post is also available in: Português (Portuguese (Brazil))

Hi guys,

Hope to you are good!

I am working in a project to migrate some databases to Oracle RAC, and we decided to move to the RAC Databases using DataGuard approach, since a Switchover usually finishes really fast and the downtime window it will be also not longer.

Well, as part of DataGuard/Standby creation process, we started a Duplicate for one of the databases (54TB size).

Since it was a Pre-Production environment and to avoid some impact, we decided to use 6 channels in target and 6 channels in auxiliary database:

connect target sysmigration/password@SOURCE_DB
connect target sysmigration/password@NEW_RAC_DB
run {
allocate auxiliary channel sby1 type disk;
allocate auxiliary channel sby2 type disk;
allocate auxiliary channel sby3 type disk;
allocate auxiliary channel sby4 type disk;
allocate auxiliary channel sby5 type disk;
allocate auxiliary channel sby6 type disk;
allocate channel pmy1 type disk;
allocate channel pmy2 type disk;
allocate channel pmy3 type disk;
allocate channel pmy4 type disk;
allocate channel pmy5 type disk;
allocate channel pmy6 type disk;
duplicate target database for standby from active database;
}

Well,

Everything started fine, but, after some hours, we just verified that for some channels, after a datafile finish the copy/restore, the channel did not start a copy of a new datafile. Instead of that, we had some sessions waiting for “remote db file read” (or sql*net vector data to client) event with time increasing with no expectations to decrease in the new database. In the source database the event was “remote db file write”.

Well, I found a Metalink (MyOracle Support) Note:

RMAN Hangs When Creating Standby DB As Duplicate From Active Primary (Doc ID 2512075.1)

For this note, the Cause is:

Issue not identified.

The Solution is:

Upgrade Linux kernel to latest version resolved the issue.

Well, for this customer we are using RHEL 7.5 since their backup software agent did not work well with RHEL 7.6.   Also, I cannot push them to upgrade since we have no confirmation with Oracle Support about the cause (Issue not identified).

I also found another note:

RMAN Duplicate hangs when using DCD and TCPS (Doc ID 1676197.1)

The Cause is:

DCD + TCPS

The Solution is:

Disable DCD.

 

For this environment customer is using DCD (Dead Connection Detection), but not TCPS.

We decided to disable DCD and restart Duplicate process, which fix the issue and we used all the channels allocated!

You can read more about it here: Dead Connection Detection – Oracle

 

Hope this helps.

 

Peace!

 

Vinicius