A few years ago my home was nicked by lighting. While the home only had a 1cm hole in the rain gutter, every piece of electronics in the home died. So I am rather diligent on backups, so I thought.
Just lost a drive hosting my 2 TB project with millions of files I've been working on for 2 years and discovered my backups have some silent failures (some files are not readable others empty which should not be). So much for being diligent by manually backing up to dropbox and google drive, and also backing up onto a number of different detached for backup hard drives, not to mention UPS. Fortunately, I have soo many backups I will be able to track down each and every file before the corruption, but it's a pain.
It turns out that uncontrol control core dumps of my development PC can do nasty things to files on Ext4 and Fat like changing file content. Pushing these files blindly into various backups just spreads the cancer.
Now motivated to get onto Truenas with ZFS to avoid a repeat of this nightmare. While I work on recovering my lost files across, I ordered some HW. With some luck, I will finish the file restoration when the HW arrives. I could use some advice on a number of things:
1) I first considered and rejected the idea of running Truenas in a VM on my development system and doing PCI passthrough to the VM so Truenas could manage the disks directly. The development PC has massive memory and cores, but I rejected this idea because of my development uses a bunch of sketchy SW packages running inside docker and somehow these manage to hard crash the host a few times over the past 2 years. Uncontrolled crashes of the host for Truenas is just asking for trouble. Is this reasonable logic or is Truenas resilient with uncontrolled crashes of the host? A single monster server would be really nice.
2) I purchased a used Dell PowerEdge T320 Server Intel Xeon E5-2430 2.20GHz 3x500GB 16GB RAM Win 10 and qty 5 of WL 4TB 64MB Cache 7200RPM (Enterprise Grade) SATA 6Gb/s 3.5" Internal Hard Drive. Is it as simple as swapping out the HD and boot with the installer for Truenas or am I missing anything?
I am assuming I can disable the built-in raid and let Truenas deal with the disks directly.
I am assuming this dinosaur PC has at least 1G ethernet.
I am assuming that Truenas can be configured to backup to dropbox and/or google drive.
Anything else likely to be missing or troublesome?
3) A bulk of the files I will now consider the truenas copy as master and push these to the development PC and to Dropbox/Google Drive. These are safe from corruption because of ZFS build-in checksum. Other files that are mastered on the development pc and backup to Truenas may get corrupted over time by the development system. So I need to develop an integrity check in my app, for timely detection of corruption. When corruption is detected by my app, I need to back up to the previous versions. The millions of files may make this less timely than I like. Does Truenas offer a way to pick a file and view its versions or checksum over time? For example: I know files /a/b/c/d/e/f.data is corrupt, do I need to check snapshot by snapshot or does Truenas offer a way to jump to the snapshot with the previous changed version of a specific file?
I am a complete newbie to Truenas, so please feel free to correct my terminology and offer any insight that may help me.
Regards,
John
Just lost a drive hosting my 2 TB project with millions of files I've been working on for 2 years and discovered my backups have some silent failures (some files are not readable others empty which should not be). So much for being diligent by manually backing up to dropbox and google drive, and also backing up onto a number of different detached for backup hard drives, not to mention UPS. Fortunately, I have soo many backups I will be able to track down each and every file before the corruption, but it's a pain.
It turns out that uncontrol control core dumps of my development PC can do nasty things to files on Ext4 and Fat like changing file content. Pushing these files blindly into various backups just spreads the cancer.
Now motivated to get onto Truenas with ZFS to avoid a repeat of this nightmare. While I work on recovering my lost files across, I ordered some HW. With some luck, I will finish the file restoration when the HW arrives. I could use some advice on a number of things:
1) I first considered and rejected the idea of running Truenas in a VM on my development system and doing PCI passthrough to the VM so Truenas could manage the disks directly. The development PC has massive memory and cores, but I rejected this idea because of my development uses a bunch of sketchy SW packages running inside docker and somehow these manage to hard crash the host a few times over the past 2 years. Uncontrolled crashes of the host for Truenas is just asking for trouble. Is this reasonable logic or is Truenas resilient with uncontrolled crashes of the host? A single monster server would be really nice.
2) I purchased a used Dell PowerEdge T320 Server Intel Xeon E5-2430 2.20GHz 3x500GB 16GB RAM Win 10 and qty 5 of WL 4TB 64MB Cache 7200RPM (Enterprise Grade) SATA 6Gb/s 3.5" Internal Hard Drive. Is it as simple as swapping out the HD and boot with the installer for Truenas or am I missing anything?
I am assuming I can disable the built-in raid and let Truenas deal with the disks directly.
I am assuming this dinosaur PC has at least 1G ethernet.
I am assuming that Truenas can be configured to backup to dropbox and/or google drive.
Anything else likely to be missing or troublesome?
3) A bulk of the files I will now consider the truenas copy as master and push these to the development PC and to Dropbox/Google Drive. These are safe from corruption because of ZFS build-in checksum. Other files that are mastered on the development pc and backup to Truenas may get corrupted over time by the development system. So I need to develop an integrity check in my app, for timely detection of corruption. When corruption is detected by my app, I need to back up to the previous versions. The millions of files may make this less timely than I like. Does Truenas offer a way to pick a file and view its versions or checksum over time? For example: I know files /a/b/c/d/e/f.data is corrupt, do I need to check snapshot by snapshot or does Truenas offer a way to jump to the snapshot with the previous changed version of a specific file?
I am a complete newbie to Truenas, so please feel free to correct my terminology and offer any insight that may help me.
Regards,
John