Question regarding Log Shipping

  • Hello

    I just have a quick question in regards to Log Shipping. Short story short, we are thinking of using LS for disaster recovery. I have successfully set up LS and it works fine.

    My question is, we currently have transaction log backups performing on the primary databases every 6 hours. I do not use the Truncate parameter in the backup log command.

    I'm noticing that my LS gets out of sync around 6pm (which is one of the Tran Log Backups) and ends up failing.

    Would this be the reason why? I would have thought by not truncating the log I could continue to back up the transactions to my primary server disk, and ship out the logs to the secondary.

    It's obvious that I'm missing something and would appreciate any feedback and knowledge transfer!

    Thanks

  • No that is not the reason.

    Whenever the log is backed up, it's automatically truncated. You need to check out why the job failed? there can be many reasons for failure. check out for log shipping job history to find out what caused failure.



    Pradeep Singh

  • Ok, so if the log is truncated every time....

    My log shipping runs every 15 minutes.

    So at 5:47pm I ship off my logs to the secondary server.

    At 6pm I back up the transaction log to disk on primary server.

    Wouldn't I ruin the sequence # for the transactions by my next log shipping backup at 6:02?

    This is the error:

    has restore threshold of 240 minutes and is out of sync.

    I will try to test a few more things out, but was just curious for thoughts.

    The copy job and every thing else works just fine. The size of the logs are extremely small (this is about a 1gb db). I suppose it could be a issue between the two machines.

  • What do you mean by it's out of sync? Is there an error? Or is there something you notice?

    You should run log backups to new files each time, without any parameters needed, and then restore those in sequence on the standby server.

    6 hours doesn't seem good for DR. That means you can lose up to 6 hours worth of data. Most people running log shipping for DR are backing up every 5-15 minutes.

  • We have two processes right now;

    A MPlan Transaction log backup that runs every 6 hours (we are more of a DW env, not OLTP). These backups are stored on disk along with our full and diff backups.

    Log Shipping that runs every 15 minutes. These logs are shipped to our secondary server, or DR site. We are testing out of LS is a feasible method (as opposed to Replication or San Replication).

    The exact error message that I show on the secondary server Log file is as follows:

    "2009-09-17 08:36:00.94 spid58 The log shipping secondary database XXXXXXX.XXXXXX has restore threshold of 240 minutes and is out of sync. No restore was performed for 889 minutes. Restored latency is 2 minutes. Check agent log and logshipping monitor information"

    The log shipping was set up, and working just fine until 6pm last night. At 6pm our MPlan Transaction log job kicked off and backed up the Tlog to disk.

    The last log shipping restore job completed at 5:47pm.

    LS restore job stopped working after that.

    My thought was if the TLog backup cleared out the log at 6pm, then is it possible the transaction sequence #'s were also cleared out, therefore creating a out of sync issue with the next LS backup.

    I have my alert threshold set to around 2/3 hours, and there doesn't appear to be any hang ups over the network with the copy job (as the copy job is still executing without issues).

  • If you have setup log shipping, you'll need to remove transaction log backup from your maintenance plan or else the log chain will break and your log shipping will fail.



    Pradeep Singh

  • Pradeep is correct. EVERY log backup needs to go to the secondary to be restored. If you keep the maintenance plan, you'll have to manually ensure those backups get moved and restored on the secondary.

    I'd recommend you remove them for the log shipped dbs.

    If you're OK with 6 hours, that's fine. Just be sure you account for data loss during that time. If an extra load or processing occurs, is that a big deal? It doesn't necessarily hurt you to log ship every hour. If nothing happens, the backups will be tiny.

Viewing 7 posts - 1 through 6 (of 6 total)

You must be logged in to reply to this topic. Login to reply