Comment Re:The tests don't show everything (Score 1) 468
Nor do they mention the days required to rebuild a RieserFS system when it gets corrupted or the day plus to even get the data off before it crashed entirely. Nor do they mention how to backup the data so you can do a recovery in the event of a hardware failure.
Real world experience tells me that RiserFS is not ready for production on my servers. I had a 2TB server running the backend mail spool for our campus, it developed severe database corruption in the ReiserFS as a result of the big blackout last summer. It took 29+ hours to remove the files that were on the system to a new system. It was also taking forever to do any kind of backup on it (tar was the only available option).
The new system we migrated the data to eventually developed an even nastier problem where just reading certain files would kernel panic the entire system, leaving us with more than an hour of down time while the system came back up and checked the ReiserFS structures and raids.
We eventually migrated to a third system where we now use dozens of individual EXT3 file systems under LVM. We have regained the ability to do a full level 0 dump in under 4 hours compared to the 24 hrs it took to save less data on a ReiserFS (when it was working properly) if we could even get the backups to complete.
And yes, all three systems were in fact identical 2TB RaidZones running their RH7.1 based product so it is an apples to apples comparison between EXT3 and ReiserFS.
The Linux LVM system is a dream from a backup perspective, as long as you aren't running RH Enterprise Advanced 2.1 with a new kernel to support LVM.
I won't use ReiserFS for anything more important than /tmp until there is a dump system available for it. I also won't use RH unless absolutely required as the only possible solution that the vendor will even talk to us about. Otherwise it is strictly Debian Sarge which doesn't ever let me down.
Real world experience tells me that RiserFS is not ready for production on my servers. I had a 2TB server running the backend mail spool for our campus, it developed severe database corruption in the ReiserFS as a result of the big blackout last summer. It took 29+ hours to remove the files that were on the system to a new system. It was also taking forever to do any kind of backup on it (tar was the only available option).
The new system we migrated the data to eventually developed an even nastier problem where just reading certain files would kernel panic the entire system, leaving us with more than an hour of down time while the system came back up and checked the ReiserFS structures and raids.
We eventually migrated to a third system where we now use dozens of individual EXT3 file systems under LVM. We have regained the ability to do a full level 0 dump in under 4 hours compared to the 24 hrs it took to save less data on a ReiserFS (when it was working properly) if we could even get the backups to complete.
And yes, all three systems were in fact identical 2TB RaidZones running their RH7.1 based product so it is an apples to apples comparison between EXT3 and ReiserFS.
The Linux LVM system is a dream from a backup perspective, as long as you aren't running RH Enterprise Advanced 2.1 with a new kernel to support LVM.
I won't use ReiserFS for anything more important than
