[gold-users] Question on the g_transaction_log table.
ingmar at ucar.edu
Thu May 26 11:03:09 MDT 2011
Thanks for the info. I now know there is a possibility the
transaction_log table could be written to by Gold.
On 05/26/2011 10:43 AM, Scott Jackson wrote:
> Well, the reason is that the database subroutines which checkpoint to the log are update, delete and undelete. And since these calls are never normally performed against the transaction object, no entries ever go into the transaction log.
> However, your question has caused me to wonder if it is a mistake that insert is not checkpointing to the journal. I will have to examine this and also evaluate the impact to performance if we were to checkpoint all transaction inserts. I think we would probably want to exclude the transaction log from insert checkpoints anyhow.
> Thanks for bringing this up. I will try to examine this in the last phase of the next gen Gold development actions.
> ----- Original Message -----
>> From: "ingmar"<ingmar at ucar.edu>
>> To: "Gold Users Mailing List"<gold-users at supercluster.org>, "Rob Perry"<rperry at ucar.edu>
>> Sent: Thursday, May 26, 2011 9:16:45 AM
>> Subject: [gold-users] Question on the g_transaction_log table.
>> I notice in both of our testing installations of Gold 18.104.22.168, that
>> g_transaction_log table is empty while the g_transaction table has
>> entries. Is the g_transaction_log table not being used by the
>> and is just there to conform with the basic set up of the other Gold
>> tables always having a log table associated?
>> -Ingmar Thompson NCAR
>> gold-users mailing list
>> gold-users at supercluster.org
> gold-users mailing list
> gold-users at supercluster.org
More information about the gold-users