Hi,

We are running currently a database with FirebirdSS-2.1.2.18118-0 at Linux box with 8 cpu cores and 8G RAM.

Server platform: 2.6.18-164.6.1.el5 #1 SMP Tue Nov 3 16:12:36 EST 2009 x86_64 x86_64 x86_64 GNU/Linux (CentOS 5.4)

Database has quite a lot of traffic (intensive inserts and updates), file size is at the moment approx 1.6G

Occasionally we are getting situations where Firebird is not available anymore to respond to requests, I think there might be a deadlock or similar at the database. If I restart all applications which are using the database it will not help. Only trick which helps is to restart the Firebird instance. After that everything works fine.

It's a production box, so I haven't had a possibility (yet) for searching locks with fb_lock_print -tool - on the other hand I'm quite inexperienced with that, whatsoever my question for you is that is this a bug when we can put Firebird to a state that it has to be restarted? If this is all about deadlocks - is there any way to release them without dbengine restart, which I believe can cause data corruption??

BTW. This has been seen at least from 2.0.3 to 2.1.2 (In some other environments I think also with 2.1.3, but not 100% sure). A Debug version would perhaps be nice, but it's not possible at this production environment: At firebird.log -file there's nothing ...

Thanks for you help!

-Pasi


[Non-text portions of this message have been removed]