On 14/07/2016 8:19 p.m., Roy Lambert wrote:
>
> My first thought reading your paragraph above was "there's two servers being run".
>
Pretty sure not 2 servers

> I don't know if that's possible or not. Was it you that's doing something with failover servers?
>
Yes, we have a failover server set up for our cloud server.  The cloud
server is just used to authenticate users (username/password/location of
"local" server/have you paid your subs?).  We have a failover for this
server.

The problem is in the "local" server on the user site.  This is where
the backups are made according to a user selected schedule and
optionally copied to the cloud server (like a DropBox).  I'm sure we
don't have multiple local servers at the user site.

> For that to happen if your initial thoughts were right (and I also think Raul is right
> so your thoughts are wrong
AGAIN!
that would mean that either the Friday backup waited
> until its correct timeslot on the Monday (that would indicate Tim went really
> weird - hmm unlikely)  or you have another backup running.
>
> Did the user happen to record the time the backups happened?

Unfortunately, the user deleted the excess backups before telling me the
problem.
>
> That say's to me "SUSPECT THE USER" - always a good start point!
>
Agreed, but in this case I can see the schedule and JOBs created and
can't think of a way to blame the user
>
> If you're still having trouble then I suggest serialising the backup jobs with a unique file name
>
Dunnit! And the file name incorporates the DB name and date time of the
start of the time of the job.
e.g. "DB1005-2016-07-02_2300.EDBBkp"

I think I'll have to shelf this one until it happens again and hopefully
no one destroys the evidence before telling me about it.

Cheers and Thanks

Jeff

On 14/07/2016 11:22 p.m., Tim Young [Elevate Software] wrote:
> Jeff,
>
> << What is supposed to happen if someone powers off the server before 5pm on the Friday - looking forward to a good weekend? >>
>
> Raul's answer is correct - the EDB Server will only run jobs that are valid to schedule for running *right now*.  If the server is down during the run window, then the job will simply be missed.
>
> Tim Young
> Elevate Software
> www.elevatesoft.com
>

Thanks Tim for clarifying that and thanks to Raul for being correct.

Cheers

Jeff

Jeff


Another thought just occured to me. If the user misremembered timestamps, or got the time right but the date wrong, it could be that something prevented the backup being deleted as it should be. AV or the OS hanging on to it.

I know my backup job doesn't check to see it the delete worked and I do, occasionally, "loose" a backup because the job couldn't overwrite the previous file.

Now the thought has made its way into my brain I'm going to have a look at how I can handle that.

Roy Lambert

Jeff

I am barging in at the end of this thread and I haven't read all the comments so apologies if I say something that has already been said.

You note that "multiple backups are being made every 2 minutes which is roughly how long a backup takes to run"

It sounds to me as though your JOB might be generating an exception prior to being fully completed. In this situation it is possible for the server to assume the JOB is unfinished and rerun it (generating a further error, generating a further rerun ...)

The EDB system EventLog is a really useful place to query to check for this type of thing, if it is showing
"JOB started at <date>"
"Exception ... .
"JOB started at <date>"
"Exception ... .
"JOB started at <date>"
"Exception ... .

You will know what is going on.

Adam


>I am barging in at the end of this thread and I haven't read all the comments so apologies if I say something that has already been said.

It hasn't so you aren't


>It sounds to me as though your JOB might be generating an exception prior to being fully completed. In this situation it is possible for the server to assume the JOB is unfinished and rerun it (generating a further error, generating a further rerun ...)

Very good suggestion. I tend to ignore the system logs - more like forget they exist since I need to look at them so often

Roy

Roy,

<< EDBManager says 2.24b2 and just checked the server - ditto - but the job would have been set before that, and the catalog is positively antique.

Do I need to delete the job & recreate it? >>

No, the NextRun column is calculated on the fly for the system information table.  Send me over your configuration and I'll see what the issue is.

Tim Young
Elevate Software
www.elevatesoft.com

On 15/07/2016 6:17 p.m., Roy Lambert wrote:
> Jeff
>
>
> Another thought just occured to me. If the user misremembered timestamps, or got the time right but the date wrong, it could be that something prevented the backup being deleted as it should be. AV or the OS hanging on to it.
>
> I know my backup job doesn't check to see it the delete worked and I do, occasionally, "loose" a backup because the job couldn't overwrite the previous file.
>
> Now the thought has made its way into my brain I'm going to have a look at how I can handle that.
>
> Roy Lambert
>

Hi Roy

This isn't an issue for me as I have a unique file name for each backup.
 What is a issue is auto deleting old backups which is something I have
to add to the TO DO list.

Cheers

Jeff

On 15/07/2016 8:14 p.m., Adam Brett wrote:
>
>
> The EDB system EventLog is a really useful place to query to check for this type of thing, if it is showing
> "JOB started at <date>"
> "Exception ... .
> "JOB started at <date>"
> "Exception ... .
> "JOB started at <date>"
> "Exception ... .
>
> You will know what is going on.
>

Thanks Adam

I haven't looked the EventLog - ever!  I'll have to get back the manual!

Cheers

Jeff

Per email:

The issue was that Roy was using a local session and not a remote session connected to the EDB Server.  When you use a local session, you cannot get a next run value because only the EDB Server is capable of executing jobs.

Tim Young
Elevate Software
www.elevatesoft.com