All replication publications suddenly disappear

[theresatan]

Hi:

Did you have the experience that publications and their related jobs suddenly disappear?

I had a transactional pull replication between two servers, and distributor and subscriber in the same box; publisher was an AA cluster server.
The schedule of pulling was once a week.

Last Friday morning everything was working perfectly, but I found all publications and their related jobs disappear around 5pm

I checked with the sysadmin and he said nothing was changed on that period.

Did you have any idea about it?

Did you have any idea that I can prevent this happen again or any alert I can set up to monitor who drop it, when it was dropped and so on?

It is a SQL 2000 box and system table doest not take trigger.

Thanks in advance

Theresa

 

[Catadmin]

I'm betting your problem is related to the cluster. If replication isn't set up as a resource in the cluster administrator, it's entirely possible that the publications would disappear when the server failed over.

We've run into problems like this with SSIS and network shares.

Anyway, check to see if the cluster failed over or if someone removed the server from the cluster somehow. Or better yet, check to see if some logical drives were removed from the server. I doubt it's the last one, but you never know.



Catadmin - MCDBA, MCSA
"No, no. Yes. No, I tried that. Yes, both ways. No, I don't know. No again. Are there any more questions?"
-- Xena, "Been There, Done That"

 

[theresatan]

Catadmin:

There was no failed over at that time period - I just checked.

No one claimed done anything on it, no logical drive lost.

I can not find any special instruction for configuring cluster server as publisher only. Do you have have any on hand?

Thanks for your quick response.

Theresa

 

[ptheriault]

Is it possible that the subscription was set to expire. I think you can configure the publication to delete itself if the subscribtion expires.

- Paul
- If at first you don't succeed, find out if the loser gets anything.

 

[Catadmin]

I can not find any special instruction for configuring cluster server as publisher only. Do you have have any on hand?

That's not quite what I meant by my cluster administrator comment. There is a tool on each server of the cluster (or should be) called Cluster Administrator. This is used to allocate resources to the cluster so when a failover occurs, you don't lose any of your settings and everything moves over to the new server correctly.

Not having used replication on a clustered server, I'm not sure what you would need to do to make sure Rep moves over during a failover, but you might check with the person (people) who installed your clustered servers and maintain them. BTW, if your cluster didn't failover during the time frame, this probably isn't your issue, but it wouldn't hurt to check up on this just in case something like this happens in the future.

Check Paul's suggestion. See if the subscriptions expired or check your Rep documentation to see if the publications expired.



Catadmin - MCDBA, MCSA
"No, no. Yes. No, I tried that. Yes, both ways. No, I don't know. No again. Are there any more questions?"
-- Xena, "Been There, Done That"