Subject Re: INET/inet_error: read errno = 9 messages + Server becomes unresponsive
Author andreas_kc
> Is your database running any better on Windows 2008? What version of Linux were you running?

For Linux we were using Fedora 7 and CentOS 5. Both had the same issues with crashing and unresponsiveness. On the performance of our database on 2008, the jury is still out. It did not crash today or yesterday which is good, but in some queries performance is awful.

Restore speed on Windows 2008 is completely night and day. It is so good, I'm completely pissed at myself for not trying it earlier. I have had at least 20 weekends so far this year that I have wasted on backup and restores because the entire process takes 36 hours for a 60GB database on Linux. On Windows I could probably do a backup/restore overnight as it takes 6-8 hours for a complete backup and restore with identical hardware!! I'm trying to convince myself I can live with the crashes as long as the backup window is this short.

Btw, I am running Classic Server on Windows where I had been using SuperServer on Linux. This may make some difference. From what I can tell on Linux, Firebird SS would create too many threads (280ish), stall out, become unresponsive and/or crash. Which I don't understand this issue. Web servers handle thousands of connections (granted this is a different type of network service that does not have to sync writing and transactional control) with many that probably don't end gracefully. Why is it Firebird seems so finicky with Network conditions to the point is crashes and corrupts the database?

I'm starting to wonder if our Cisco firewall is causing much of these errors, but I have no way of knowing. It is a little over a year old and probably the only piece of hardware that has not been replaced.

If this doesn't work, I will likely be trying to get a support contract to trace this issue down. I have had too many wasted weekends and now a bunch of pissed off employees.