Exchange CPS and Tape Scheduling Tips

[Steve_Mullenix]

Hello,

I was wondering if anyone could give me a few tips on scheduling my monthly tape backups with my Exchange CPS job.  Since the CPS always runs, there must be times where it overlaps with other jobs.  However, it seems that if the recovery point jobs (every 8hrs) kick off during a copy job to tape it hoses the CPS protection until I stop the job, restart all services, and then start the job over from scratch (ie. Full must run).

Problem is that this generally leaves a week where I do not have reliable backups since the Full job can not run during the week due to time limitations.

Should I be manually pausing the CPS job prior to the tape job?  What exactly is the right way to stop/hold and then restart CPS jobs?  It seems like I'm always running into the same scenarios with CPS:

1) Recovery point jobs never run though it does appear to be log shipping

2) Recovery jobs are run but I can't view mailboxes via restore

3) Other backups and even potentially the DB Maint. Jobs seem to disrupt it.

Any advice appreciated,

Steve

[Dean_Bliss]

Steve,

Maybe if I explain what Backup Exec and CPS are doing it may help you to configure your jobs accordingly.

First - Backup Exec takes a full backup of your Exchange server (to disk with GRT, I presume - as this is how CPS jobs need to be configured)

After the Full backup, CPS continuously replicates the Exchange Transaction Logs over to the Backup Exec server and places them into the B2D folder (.Exchange).

When the Recovery Point job runs - Backup Exec simply takes the Full backup and the replicated logs and plays back the logs to the edb file and creates a new IMG folder with this combined backup. All of this is done on the Backup Server. Once the Recovery point is created BE will send a command to the Exchange server to flush the logs that were backed up.

There is nothing further done with the CPS job - CPS just keeps replicating the logs as they exist on the Exchange server - when they get flushed - CPS simply replicates the new logs.

Is there an alert in the CPS console as to why CPS job stops running?

[Steve_Mullenix]

Hi Dean,

Thanks for the response.  The CPS console never shows an error message.  The log shipping always works actually (I've never seen it not ship logs correctly), it is the creation of recovery points that seems to be the main area that creates issues.  I thought it was supposed to be completely local at the RP stage as you say.  That is why I'm so confused as to why it seems to break as soon an another job kicks off, especially an Exchange copy job.

Is opening the recovery panel and attempting to tree down to the mailbox level, even if a restore job is not run enough to trigger this?

If I put the job on hold and then restart it I can pretty much guarantee that no RP jobs will run until the next full runs, forced or not.

One part I did not understand correctly was that the RP jobs are responsible for controlling log growth.  This occurs without incident which is strange because on the BE side it appears that the jobs either haven't run or contain no data.  So if I look at the folder level it looks ok to me, but when I try and verify this via the BE restore console it looks like Exchange is in a precarious spot.

Thanks,

Steve

Message Edited by Steve Mullenix on 06-11-200705:11 PM

[Dean_Bliss]

Steve,

Both the Recovery Point jobs and the Exchange Copy jobs use the VirtFile driver on the Backup Exec server. So that's where the conflict lies. Anyway you can schedule the copy job during a time period that the Recovery job isn't running?

Not sure I understand what you said about log growth. Do the logs get flushed on your Exchange server? - but in BE you don't see a restore point (or individual items) and so it appears that a backup hasn't run?

[Steve_Mullenix]

Any tips on troubleshooting the VirtFile driver?  If each job uses this driver then it seems to me that there could be conflicts if two jobs are accessing it at the same time.  I have not yet run a recovery job, just looked at the recovery selections in the treeview. 

As far as the log growth goes, that seems to sum it up quite well.  Logs are getting flushed but I either see no or empty recovery points.

Could this just be a quirk with the recovery browsing itself?  If push came to shove is it possible to recover items directly from the img folders?

Message Edited by Steve Mullenix on 06-12-200710:49 AM