Notes/Domino 6 and 7 Forum : RE: Are you sure you don't have bad hardware?

>Are you sure you don't have bad hardware

At one point we did have a bad stick of ram which was causing issues. We swapped that out and our linux kernel panics stopped. However, the domino & db problems continued. We also saw the corruption across multiple servers - at least 3. I had a Windows 2000 server (dell server hardware) which was perfectly stable for at least 2 years crash just days after deploying the client's database. We then swapped out the Windows 2000 server (older P2 xeons) with a custom built high-end Athlon server running linux - same problem there. We also built another new linux server for testing and future deployment when the failover code we wrote is done, and it crashed that server once or twice. Strangely, however, the one server that was the most stable was the Windows NT 4 workstation box. The Windows NT 4 workstation box was an ancient P2 300 machine which was the company's first server. We dug it out from under a pile of dust bunnies and fired it up, so at least the client could operate on a server without affecting the other clients if it crashed. The windows NT 4 box was "mostly" stable, however, it did crash once or twice over a 1.5 month span. However, this was much less than the almost daily crashes we were having on the Linux & Windows 2000 boxes.

I would have thought the NT 4 box would crash the most, go figure!! :-)

>First thing I'd do if I were you would be to put the applications on a known-stable Windows server and reimport the data if you can.

Unfortinately, the clients have now been using this database (in a limited fashion) for two months. Reimporting would loose all of those changes, so unfortinately it's not an option at this time. We're currently working with the client to schedule a timeframe when we can built them a brand new database, run an agent to copy over their data (only the stuff we're sure should be there, in case theres a weird nasty document somewhere), and replace the design. Hopefully we'll be able to do this next weekend.

Most of the problems with the client's database have been resolved, its back to normal for the most part, immediately after we deleted the orphaned agent data notes.

We are currently working on a plan to deploy a new server on Windows 2000 with the same hardware as the linux one. The two will be deployed in a cluster using the software that I wrote. We want to compare the Windows 2000 and Linux box to see which one has better performance, and to see if any of the problems we've been having are only on one of the platforms. Hopefully the server will be purchased & ready around the end of the month.... gotta get through the financial department first <grin>

>I've never seen that lost agent note error you mentioned.

Download the "compagnt" tool from Lotus and run it on one of your databases with the -f command. You may be surprised.... I didn't expect to find any.. I found over 300 in some databases, and over 1,000 in one of our templates! I really have no idea why they caused the server problems with that one client's database, but only slowed down all of the others. Very strange....

> And why are you getting it if your application isn't changing?

Our company's service model is tuned for frequent updates. We fix and deploy bugs typically within a 1 week turnaround time, or less.

> How much data is in the databases and what do your views look like?
>Sounds huge if you sucked them in from a SQL database.

The one client database we imported from the two SQL databases (and one notes database) is approx 2.5 gigs after replication (no views built). After average use, the database size grows to about 5 gb. I believe there are around 400,000 documents.

Notes/Domino 6 and 7 Forum

Document Options

Search this forum

Forum views and search

Member Tools

RSS Feeds