CCA backup

[Vic]

Hi,

I am having a major dilemma here with DataSourcer: how do I prevent data loss when datasourcer goes down?

Well, the obvious solution I thought would be having a second datasourcer operating in standby, by then, we still would miss some events, wouldn't we? Or would we?

DataSourcer does not operate on events, but instead queries statisrtics from statserver. So, as long as statserver is up, we could query the statistics for a particular DN even after restating a datasourcer. It would seem like as long as time profile is longer than datasourcer downtime, we would be able to retrieve statistics with no problem.

But, then there is a really good question: when you query a statistics, who decides on the starting time for that range? Is it the query time minus time range?

Or is it time range defined in STATSERVER?
I know it depends on definition in STATSERVER, and unless it is a sliding window stat, we would get 12:0012:15 statistic when queried at 12:20. So, then if we datasourcer goes down and then starts again, wouldn't it query the data at 12:21 and get the same statistic? Thus I think we should be in a clear... But then again, why is it not like this?

I know it is a very strange question, but I was wondering if someone could help me in trying to figure it out.

Thanks,
Vic

[AR]

Hi,

I'm not sure if this is what you are lokking for:

DataSourcer only collects statics from StatServer 12:00, 12: 15 etc. This means that if DataSourcer has successfully written the 12:00 chunks, goes down at 12:03 and is restarted at 12:14 no stat is lost. DS will collect stat again at 12:15.

A restart of StatServer between 12:00 and 12:15 means that stat is lost.

[Tony]

I don't believe Datasourcer can be run in a backup mode, I checked with Genesys on this before and I believe it hasn't been developed. The other respondee was correct as the Datasourcer can be stopped anywhere between the 15 minute intervals and no data will be lost. Let me know if you were able to find anything else out regarding the backup version.

[Vic]

I was thinking of the same thing, but there is something I want to check with all of you:

doesn't datasourcer wait for one timerange before attempting to collect statistics after it starts up?
In other words,

data sourcer crashed at 12:16, so 12:0012:15 minute chunk has not been collected. If we restart datasourcer at 12:17, would it not wait until 12:32 and only then collect 12:1512:30 chunk? So we would be losing 12:0015 chunk?

Anyone can check this?

[CCA Guru]

This is correct. Genesys have not included any redudnacy within Datasourcer but there is a workaround which you can use. You need to start 2 DS'a both using the same application name. One will start as normal but the other will report 'application violation' errors. When the 1st ones stops the second one will kick in. ONly issue ensure emergency file is empty.

[Dana Lee]

We have spoken with Genesys on this subject and were told that we could have two DataSourcers running each writing to it's own database. In the event one of the DataSourcers goes down, the other one will have written the data to it's database. The "pain" in this solution, is that you have to manually pull the data out of the backup database and put it into the other database.

We have not tried this.

Thanks,
Dana Lee

[Grzegorz Ostrowski]

I have used this solution some time before with 6.1 version and got not very good results (sometimes chunks were missed in one or the second database). The solution you can use to avoid manual migration of data between 2 databases (each one for each Datasourcer) is (Oracle case) to define database links and put views combining both sources automatically. Then you can use those views for reporting.