So I managed to get someone from the Redhat forums to have a look at the bug report.
In a nutshell this isn't just localised to zfs - it applies to all applications that exert extremely heavy i/o operations when using the nvme storage driver. Redhat have a fix and that fix is in QA for RHEL 7.5. Apparently the fix is being back ported to kernel-3.10.0-514.30.1.el7 as well - I've asked for the patches so that I can apply them to the source rpm myself; haven't received a response yet.
This leaves me in a bit of predicament because I don't think I can trust nvme in production given this issue. If RH don't provide the patches, then the only way forward I see is temporarily migrating to kernel-lt from ELRepo, but I'd rather not to be honest.
Kernel Panic Advice
Re: Kernel Panic Advice
The 514-30 kernel you mention will never see CentOS as that's a Z stream kernel for 7.3. If you're lucky your issue is already fixed in 7.4's -693 kernel or RH will backport to the fix to that.
The future appears to be RHEL or Debian. I think I'm going Debian.
Info for USB installs on http://wiki.centos.org/HowTos/InstallFromUSBkey
CentOS 5 and 6 are deadest, do not use them.
Use the FAQ Luke
Info for USB installs on http://wiki.centos.org/HowTos/InstallFromUSBkey
CentOS 5 and 6 are deadest, do not use them.
Use the FAQ Luke
Re: Kernel Panic Advice
Thanks Trevor, but...
This strikes me as an incredibly hard thing to work out:
EDIT: OK so have been told by Redhat that these changes have been backported to 7.4 - great! The people I spoke with were all very helpful, but I'm still bemused by how difficult it is to track these things. I'm coming from Fedora server predominantly so I guess I'm not used to bugzilla being closed.
Guess I've got to hold tight for 7.4.
This strikes me as an incredibly hard thing to work out:
- The original bug is private - therefore no way of knowing what the changes were that fixed it.
- All the bug entries in the kernel spec file for 693 are also marked as private and therefore no way of knowing whether the items are related.
EDIT: OK so have been told by Redhat that these changes have been backported to 7.4 - great! The people I spoke with were all very helpful, but I'm still bemused by how difficult it is to track these things. I'm coming from Fedora server predominantly so I guess I'm not used to bugzilla being closed.
Guess I've got to hold tight for 7.4.