This topic has been marked solved and closed to new posts due to inactivity. We hope you'll join the conversation by posting to an open topic or starting a new one.
Hi all, Today one of my ReadyNAS 526's was uncleanly shutdown due to an error on my part... The unit was idle at the time and has come back online but I am keen to understand how an unclean shutdown affects the BTRFS filesystem that these NAS's use. Ordinarily I would run an fsck (if it were ext3/4) but with BTRFS there is no such tool. Should I be concerned about possible filesystem damage / corruption? How does BTRFS handle unclean shutdowns? Thanks...
The file system itself is designed to handle unclean shutdown cleanly (with no need for file system checking).
The software raid underneath is generally a bigger concern, as lost writes can result in a RAID array that is out of sync. If it does fall out of sync, the volume won't be mounted (so that definitely didn't happen in your case). The RAID sync vulnerability isn't filesystem dependent - it's the same with ext3 as it is with btrfs.
The file system itself is designed to handle unclean shutdown cleanly (with no need for file system checking).
The software raid underneath is generally a bigger concern, as lost writes can result in a RAID array that is out of sync. If it does fall out of sync, the volume won't be mounted (so that definitely didn't happen in your case). The RAID sync vulnerability isn't filesystem dependent - it's the same with ext3 as it is with btrfs.
Thanks Stephen... The RAID vulnerability is however quite a concern (not for this occasion but generally). Is there anything that can be done to prevent this and / or clean it up should the array become out of sync? Are Netgear likely to resolve this in the future or is it a feature limitation of the implementation?
There are event counters on each disk, which mdadm uses to detect if the array is in sync. There is a command option (--force) which tells mdadm to ignore the counters. If you use that, then the array will mount, but there will be some volume corruption (since there are lost writes).
I think the possibility of lost writes as a result of an unexpected shutdown (or crash) is inherent to RAID. I always recommend using a UPS with a NAS, as that eliminates the possibility of an unexpected power cut. Though of course a crash can create the same effect. Disabling the write cache is another way to substantially reduce the risk (with a performance cost) - though write cache controls aren't available in the current web ui.
