Date: Mon, 21 May 2018 08:58:47 +0200 From: Borja Marcos <borjam@sarenet.es> To: Michelle Sullivan <michelle@sorbs.net> Cc: Alex Aminoff <aminoff@nber.org>, freebsd-fs@freebsd.org Subject: Re: Can one remove a specific file from all snapshots? Message-ID: <4EA8DFA5-734E-4BBA-A267-F6C76FE44765@sarenet.es> In-Reply-To: <4b59e335-9140-5d29-4177-00bc33d65980@sorbs.net> References: <38792b3b-04db-0c7f-c9a4-3f55b908a427@nber.org> <4b59e335-9140-5d29-4177-00bc33d65980@sorbs.net>
next in thread | previous in thread | raw e-mail | index | archive | help
> On 21 May 2018, at 04:36, Michelle Sullivan <michelle@sorbs.net> = wrote: >=20 > Of course if it's making use of an appliance/existing hardware that = uses ZFS by default/design you could always do something smart like = creating a zvol for said data and formatting it with an FS type that = doesn't do copy-on-write and doesn't have that type of snapshops etc... = perhaps with a software mirror zvol if it's redundancy you're after... = perhaps as a 'this is your secure/non snapshotted' drive export tag = etc... 100's of solutions depending on the client needs. Beware, the zvol will still do copy on write, you can do snapshots of a = zvol, etc :)=20 The key in most cases is to exploit the flexibility of ZFS better. You = can do as Michelle says, or just increase your dataset granularity.=20 Do you have data that needs more snapshots with a longer retention = period? That must go to a separate dataset. Conversely, data that needs a shorter retention period? Separate = datasets. Data that might require a rollback without affecting others? (Example, a = user=E2=80=99s home directory in a file server in case of a ransomware = incident) Separate dataset And so on. I guess some still treat datasets like disk partition in the = old times, with a limited number per disk.=20 Cheers, Borja.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4EA8DFA5-734E-4BBA-A267-F6C76FE44765>