Hi all, Nevada build 67, USB flash voayager, ... created zpool on one of the FDISK partitions on the flash drive, zpool import export works fine, tried to take the USB stick out of the system while the pool is mounted, ..., 3 seconds, bang, kernel down, core dumped, friendly reboot on the way... it might be a faq or known problem, but it''s rather dangerous, is this being worked ON? usb stick removal should not panic the kernel, should it? thanx, Martin
zfs-discuss-bounces at opensolaris.org wrote on 07/12/2007 07:28:29 AM:> Hi all, > > Nevada build 67, USB flash voayager, ... > > created zpool on one of the FDISK partitions on the flash drive, zpool > import export works fine, > > tried to take the USB stick out of the system while the pool is mounted, > ..., 3 seconds, bang, kernel down, core dumped, friendly reboot on the > way... > > it might be a faq or known problem, but it''s rather dangerous, is this > being worked ON? usb stick removal should not panic the kernel, shouldit? I think the default behavior is that if the pool is unprotected (or at an unprotected state via redundancy failure on mirror or raidz(2)) and you lose a device the system panics. This is a known issue/bug/feature (pick one depending on your view) that has been discussed multiple times on the list. -Wade
Wade.Stuart at fallon.com wrote:> Martin.Man at sun.com wrote: >> it might be a faq or known problem, but it''s rather dangerous, is this >> being worked ON? usb stick removal should not panic the kernel, should > it? > > I think the default behavior is that if the pool is unprotected (or at an > unprotected state via redundancy failure on mirror or raidz(2)) and you > lose a device the system panics. This is a known issue/bug/feature (pick > one depending on your view) that has been discussed multiple times on the > list.discussed yes, I think I remember that, reported? being worked on?> -Wadethanx, Martin
Martin Man wrote:> Wade.Stuart at fallon.com wrote: >> Martin.Man at sun.com wrote: >>> it might be a faq or known problem, but it''s rather dangerous, is this >>> being worked ON? usb stick removal should not panic the kernel, should >> it? >> >> I think the default behavior is that if the pool is unprotected (or at an >> unprotected state via redundancy failure on mirror or raidz(2)) and you >> lose a device the system panics. This is a known issue/bug/feature (pick >> one depending on your view) that has been discussed multiple times on the >> list. > > discussed yes, I think I remember that, reported? being worked on?Martin, see: 6322646 ZFS should gracefully handle all devices failing (when writing) Menno -- Menno Lageman - Sun Microsystems - http://blogs.sun.com/menno