tsm 7.1.1 log pinned with backup node NDMP whitout TOC

[tpesselier]

my actlog are pinned every time when i do an NDMP backup (30T) whitout TOC .
Log increase 90pcrt and backup fail.
IBM told me to increase space log 128 o 512GB but i don't thiink that's the solution beacause whitout backup node NDMP i never exceeds 20 prct .
Have you any idea ? thx

 

[marclant]

I'm not sure if you know or understand what log pinning is or not. I'll explain anyway. Here's an image:


In the above, you can see T2 or transaction 2 starts, as long at T2 is still running, it's pinning the log. Any transactions that are started after T2 will consume active log space which will only be freed when T2 ends.

So with an NDMP backup, you have really long transactions, which pin the log. There are a few things you can do to reduce the impact:

• re-arrange backup schedules so there is less activity during NDMP backups, regular file backups create a lot of transactions, while NDMP long transactions, the combination results in larger active log consumption
  
• shorter transactions by creating smaller NAS volumes, which means they will backup quicker, so less log consumption because not pinned as long
• increase size of active log

 

[tpesselier]

Hi

Thank you for your reply ,where you find this information ? i am very interested.
I am agree whith you but whith less log space i could save this volume whith 6.3.4 version ,i do not understand why .
IBM told me that whitout toc the problem is fix , but no ....

 

[marclant]

The image, I just found using Google. What I explained is just how a standard database transaction log works, doesn't matter if the DB is DB2, Oracle, MSSQL, etc.

 

[tpesselier]

for information ,the txngroupmax can be set to a "backup node" , reducing the txngroupmax to avoid pinning log