ADSM-L

Re: [ADSM-L] AW: AW: Tsm replication 7.1

2014-05-19 22:00:41
Subject: Re: [ADSM-L] AW: AW: Tsm replication 7.1
From: "Prather, Wanda" <Wanda.Prather AT ICFI DOT COM>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Tue, 20 May 2014 01:57:50 +0000
From EDGE:
Different retention on source vs target server is planned for a future release.

(with all due disclaimers about how nothing is guaranteed for future 
releases...)



-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf Of 
michael.malitz AT TSMPOWERADMIN DOT COM
Sent: Monday, May 19, 2014 1:33 PM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: [ADSM-L] AW: AW: Tsm replication 7.1

Hallo Robert,  

to answer the last part of your question:

"...So if I have for a node a policy (archive) of keeping 365 days and on the 
TARGET server I just want to keep 60 days. During the first replicate node 
which policy works ? All data for 365 days will be replicate or just the data 
for 60 days ? ..."

Some Facts:

Domains are not replicated. If same domain name  exits on Target Replication 
Server (TRS), it will be used, otherwise the “standard” one will be choosen. 

Expiration is always handled by the source replication server (SRS). SRS 
informs TRS to delete an expired Object during replication process.


Case 1:  backups:

Assume you allow 4 versions on SRS and only 2 versions on TRS:

Replication will replicate all 4 versions (you can check with q occ and from 
“backups”  table on TRS).

“Half-Failover: If you change to the TRS with –forcefailover  (or with a DSM 
optfile pointing to TRS (same result)) , you can restore all 4 versions from 
TRS.

“Full-Failover”: Now, you do a   “remove replnode <node>”   on TRS :  
replication is removed (replstate=none and replmode=none)

 You still  !! can restore 4 versions on TRS:  Only if you do a new BACKUP 
(which now is possible), a “rebinding”  will be done, forcing the policies of 
the TRS to become  effective (2 versions only).

Case 2:  archives:

As long as the SRS is the owner, you will have RETVER=365 for all replications. 
But if you change ownership of your node on TRS (remove replnode <node>), than  
RETVER=60 days will become the effective value for your archives. The next 
“expire inventory” will expire and delete all archive objects older than 60 
days. 

Some background for archives:  Archives are never rebound (in terms of 
active/inactive objects), because we have no versioning. 
Archive copies remain bound to the  MANAGEMENT CLASS  specified, when you 
archive them (only if the MG does not exist anymore or the archive CP is 
deleted it will be rebinded to the default MG (or the grace period if the 
default MG is not available).

Attention:  pls check also the retention for your archive directories (may be 
you are using also ARCHMC=)


So the final message is:    keep SRS and TRS policies the SAME for a node to 
replicated!.


 rgds Michael
 
michael.malitz AT tsmpoweradmin DOT com

-----Ursprüngliche Nachricht-----
Von: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] Im Auftrag 
von Robert Ouzen
Gesendet: Donnerstag, 1. Mai 2014 07:35
An: ADSM-L AT VM.MARIST DOT EDU
Betreff: Re: AW: Tsm replication 7.1

Hi Michael

Thank you very much for the information ..... make more sense now.

Your link is very helpful too ..

I have just need another clarification   about the policy retention , if they 
are not the same in SOURCE and TARGET servers . I mean that I want to keep in 
TARGET server less versions (type backup) or last days (type archive).

To make simple in source F.S (default) - Node (default) - server (ALL_DATA)

So if I have for a node a policy (archive) of keeping 365 days and on the 
TARGET server I just want to keep 60 days. 

During the first replicate node which policy works ? All data for 365 days will 
be replicate or just the data for 60 days ?

T.I.A Best Regards

Robert Ouzen
Haifa University Israel

-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf Of 
Michael malitz
Sent: Wednesday, April 30, 2014 3:49 PM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: [ADSM-L] AW: Tsm replication 7.1

Hallo Robert,

 

interesting questions:

 

As I'm currently  documenting and testing as well the node replication feature 
(see
http://www.mm-it.at/de/TSM_Node_Replication_CMDs_Options_at_a_glance.html)

 

I try to share my current knowledge:

 

Domain Change on Target Replication Server:

 

If you move a node to a different domain on the target, replication will move 
it back

 

Means, that if I change a domain for the node on the target replication server 
AFTER a successful replication and then try another replication to the same 
node ,

the domain for the node on the target server changed back (without a
message!!!) to the original one.

 

replstate=purgedata

 

E.g. if you do:   de fi node-name * replstate=purgedata  datatype=backup,
then:

 

è Backup rule data is then set to  state "Purge Data":    à  see  q   fi
node *  f=d

è During next replication the objects will be deleted on the target server (see 
message during replication) and

è The replstate for the filespace will be set to disabled 

 

 

You can see the status of the deleted objects later on with   q replicaton
node-name  <processed=xx> f=d

 

What remains to be clarified for me:  also with "update node"  you have the 
"undocumented" option of purgedata  for the replstate, but I can not see any 
"purge data" status

 

è My be TSM development can clarify - thanks 

 

Expiration:

 

The Source server instructs target to delete objects during replication

 

rgds Michael

 

michael.malitz AT tsmpoweradmin DOT com

 

-----Ursprüngliche Nachricht-----
Von: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] Im Auftrag 
von Robert Ouzen
Gesendet: Mittwoch, 30. April 2014 11:53
An: ADSM-L AT VM.MARIST DOT EDU
Betreff: Tsm replication 7.1

 

Hi to all

 

Long time ago I test the replication feature for DRM purpose. Now I wanted to 
implant it between two TSM servers version 7.1 and O.S Windows 2008 R2 64B.

 

I have some questions:

 

 

·         The command replicate node if no domain exist in the TARGET server
as in SOURCE server  , put the node under the domain STANDARD.

 

My question is after it can I update the node to a different domain (created on 
the TARGET server but  in the SOURCE server is not exist) , and in the next 
replicate node it will go to this domain and not to STANDARD ?

 

 

·         In SOURCE server in the command "update fi" it is an option
replstate=purgedata"   I don't quite understand the purpose  ?

 

 

 

·         When replicate node  with type ARCH  have in  ARREPLRULEDEFAULT
the options :  ALL_DATA ,   ALL_DATA_HIGH_PRIORITY ,  DEFAULT ,  NONE  . I
am using ALL_DATA

 

If I have a node in the SOURCE server with archive copygroups  rentention of
360 days and in the TARGET server in the archive  coypgroups retention of the 
domain STANDARD  is 90.

 

My question what happen in the replicate node  to the TARGET server  when no 
same archive copygroups retention ? Did all data will be replicate and only 
after running an expire inventory in the TARGET server only the data for 90 
days will remain ? or already during the replicate node only the data for 90 
days with will replicate ?

 

 

 

T.I.A  Best Regards

 

 

 

Robert
<Prev in Thread] Current Thread [Next in Thread>