[Bug 236220] ZFS vnode deadlock on zfs_mknode

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=236220

Dennis Noordsij <dennis.noordsij at alumni.helsinki.fi> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dennis.noordsij at alumni.hels
                   |                            |inki.fi

--- Comment #20 from Dennis Noordsij <dennis.noordsij at alumni.helsinki.fi> ---
As an extra datapoint, I've noticed this exact same problem after upgrading to
12.0-RELEASE in combination with PostgreSQL 11.2

At first I blamed AWS EBS volume hiccups for the stuck-on-access symptoms, but
now outside of AWS (KVM based VPS with NAS) I am running into the same problem.

Basic system, a UFS boot partition and a ZFS pool with a root dataset and a
database dataset. Dedicated to running PostgreSQL, no other apps running apart
from SSH, no jails, etc. Not hitting memory/CPU limits.

Database has recordsize=8k and compression=lz4

It has so far always happened on a sunday/monday, one or twee weeks apart
(coincidentally on saturday night a basebackup and an SQL backup are taken
which would create some disk/postgresql activity).

Obviously very inconvenient. I can't reproduce it on a dev system (which has a
similar hw/sw config but of course not a real usage pattern).

Would appreciate any workarounds or suggestions.

-- 
You are receiving this mail because:
You are the assignee for the bug.