Closed devZer0 closed 5 years ago
it also happens with 0.7.12
it also happens with 0.7.12
Comparing 0.8.x with 0.7.x seems to suggest that some of the code responsible for the backtrace in your report has changed, and from a cursory reading i would have guessed this oops cannot happen in the 0.7.x release branch.
If you are actually able to reproduce this running 0.7.x please post a new backtrace.
yes, you may be right, i will re-check. i found yesterday that when downgrading the 0.8rc2 install to 0.7.12 i was curious, that before reboot i could use the pool but after reboot i could not import it due to "missing features". and that makes me think that perhaps 0.8rc2 zfs modules may not have been unloaded and 0.7.12 loaded before re-testing this issue... i did the downgrade to check if the problem is reproducible there
you are right, my last observation must be wrong. i was not able to reproduce the problem with 0.7.12 again.
Distribution Name | CentOS7 Distribution Version | CentOS Linux release 7.5.1804 (Core) Linux Kernel | 3.10.0-862.14.4.el7.x86_64 Architecture | x86_64 ZFS Version | 0.8.0-rc2 ( zfs-0.8.0-rc2.el7.x86_64 ) SPL Version | 0.8.0-rc2
Describe the problem you're observing
While playing with metadata allocation classes and filling disks/pool with some test data, one of the pool's ordinary disks faulted (using old, flaky ones), leaving kernel BUG/trace below. not sure if this has anything to do with special device/metadata separation... oh, and ignore there is double fault, the second drive faulted while resilvering...