Debugging the InterBase blogs server

The blogs server was having problems on Friday and John Kaster asked me to look at the InterBase side of things. (Note that John has a separare blog discussing debugging the blobs server). While John did give me access to the database via Remote desktop, I didn't need this and I thought I would briefly discuss how easy it was looking at what InterBase was doing.

Based on the problems being observed, it was theorized that the InterBase server had been crashing and possibly the InterBase Guardian was restarting the InterBase server, which is why reloading a page always resolved errors. I also wanted to make sure the current release of InterBase was being used.

In giving me remote desktop access, inherantly I was also told the name of the server. I was told the path to the database and the sysdba password. This gave me all the details I needed. First, I connected to the server with IBConsole, and selected Server | Properties, and selected the General tab. Here I could see the server had the current version (7.1.0.189), 2 databases were being used and the number of attachments was 19. 2 database was exactly what I expected, as one was admin.ib and the other was the database for the blogs server. If I had not been told the path to the database I could have figured it how here as well. I could also see the capabilities for the server and see they were something reasonable.

Next, I wanted to look at the InterBase.log file. So, I selected Server | View Log File from IBConsole. I saw the last time the server had restarted was the middle of night and problems had been observed long since then, so the server restarting wasn't the problem.

Next, I wanted to see what has happening with the blobs database so I connected to the database with IBConsole and ran Performance Monitor (Database | Performance Monitor). I saw 17 connections to the database, most were very recently, but 5 were hours old. While this wouldn't cause the problem at hand, it seemed odd, so I let John know about this. I checked later and they were gone, so John dealt with them somehow. These long running transactions were also another confirmation that the server had not restarted in some time. Finally, I selected the Statements tab and looked for any long running queries. There were none.

My conclusion in just a short period of time, while not even being at the server, was that InterBase was alive and kicking. The problem had to be somewhere other than the InterBase server.