18 ноября 2014 г. 11:20:21 CET, Fred Liu via illumos-discuss <[email protected]> пишет: >Sorry if it is a duplicated topic. > >For illumos-based OSes, patching ZFS means patching kernel. Solaris >used to have this possibility. >What is the current status in this aspect in illumos now? >For ZoL, zfs is just a loadable kernel module. The lively patching zfs >may have high possibility. > >In application aspect, for the ways applications own to access storage >service vary, interruption >may be incurred inevitably. > >Just a brainstorm. > >Thanks. > >Fred > > >------------------------------------------- >illumos-discuss >Archives: https://www.listbox.com/member/archive/182180/=now >RSS Feed: >https://www.listbox.com/member/archive/rss/182180/22642773-76f3f1fc >Modify Your Subscription: >https://www.listbox.com/member/?&; >Powered by Listbox: http://www.listbox.com
IMHO, since changes to the zfs module can generally involve format, caching, API, etc. changes )even if the update does not touch anything else in the OS/kernel/libc), it makes sense that all zfs storage should be unmounted and exported before reimporting it with new drivers. Hence client apps/zones should be stopped (unless they rely on a vfs like nfs, cachefs or iscsi that should make changes to storage transparent). Likely that includes to OS if the rpool is, well, a zfs pool. And then you have the fastreboot (if all works well) which runs a POST-less new kernel instantly and starts up the storage and apps in an orderly fashion. My 2c Jim -- Typos courtesy of K-9 Mail on my Samsung Android ------------------------------------------- illumos-discuss Archives: https://www.listbox.com/member/archive/182180/=now RSS Feed: https://www.listbox.com/member/archive/rss/182180/21175430-2e6923be Modify Your Subscription: https://www.listbox.com/member/?member_id=21175430&id_secret=21175430-6a77cda4 Powered by Listbox: http://www.listbox.com
