We have a two node Domino cluster (Domino only, not a Windows cluster). One server in the cluster starts up quick each time. The other always takes about one hour to start. The ‘nlogasio.exe’ process runs at 25% CPU the entire time.
The only difference between the two servers that I can see that would make a difference is that transactional logging is on a seperate partition on the first server, but not the second.
Would that cause it to be that slow? Thanks for any help you can give on this!
I would certainly suspect transaction logging. There are no other errors or indications on the console or in the logs?
Recommendations for TL is to have it on a different partition than the server itself, but what’s more important about the partition is that it’s a different drive. No sense in taking a drive and making 2 partitions, it’s the disk i/o that’s going to go through the roof logging the transactions and that shouldn’t be shared on the same drive as the server.
Why not disable TL for test run or will that really throw off your backups? If you’re clustered, are you backing up both systems? If not, why run TL on the one not being backed up? I know there are some benefits to running TL, aside form backups and restores, faster restart times after a crash for instance, but I don’t think you’re seeing that!!
No other errors on the console or the logs. Thanks for your insights into transaction logging on this server. I believe we are backing up both systems in the cluster.
We just started seeing this on one of our servers a few weeks ago. I can’t explain it by way of any environmental changes (configuration, disk, etc.) and have submitted a PMR to IBM to help investigate the cause. Perhaps there are translogging debug parms that they will advise that will expose what’s taking so long.
We’re seeing this even after a clean shutdown. Are you? My other server with the exact same config. (same drive configs, same trans logging style, etc.) starts up in about 7 seconds. I’m seeing over 45m.