zfs/cmd/zfs
Alan Somers b64afa41d5 Better control the thread pool size when mounting datasets
Ever since a10d50f999, ZFS has mounted file systems in parallel when
importing a pool.  It uses a fixed size of 512 for the thread pool.  But
since c183d164aa, it has also imported pools in parallel.  So the total
number of threads at one time is 513 * npools + 1.  That can easily
exceed the system's limit on the number of threads per process, which
will cause one or more pools to be unable to allocate any worker
threads, forcing them to fallback to slow serial mounting .  To
forestall that, manage the threadpool size in /sbin/zpool, not libzfs.
Use the same size (512), but divided by the number of pools.

This is a backwards-incompatible change to the libzfs abi.

Sponsored by: Axcient
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Wilson <gwilson@delphix.com>
Signed-off-by: Alan Somers <asomers@FreeBSD.org>
Closes #16178
2024-05-14 09:36:21 -07:00
..
Makefile.am autoconf: use include directives instead of recursing down cmd 2022-05-10 10:18:38 -07:00
zfs_iter.c libzfs: add v2 iterator interfaces 2023-04-10 11:53:02 -07:00
zfs_iter.h zfs list: Allow more fields in ZFS_ITER_SIMPLE mode 2022-12-13 17:27:54 -08:00
zfs_main.c Better control the thread pool size when mounting datasets 2024-05-14 09:36:21 -07:00
zfs_project.c Replace dead opensolaris.org license link 2022-07-11 14:16:13 -07:00
zfs_projectutil.h Replace dead opensolaris.org license link 2022-07-11 14:16:13 -07:00
zfs_util.h Replace dead opensolaris.org license link 2022-07-11 14:16:13 -07:00