Date: Fri, 23 Jan 2015 17:41:34 +0000 (UTC) From: Xin LI <delphij@FreeBSD.org> To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-stable@freebsd.org, svn-src-stable-10@freebsd.org Subject: svn commit: r277576 - stable/10/sys/cddl/contrib/opensolaris/uts/common/fs/zfs Message-ID: <201501231741.t0NHfY7u097857@svn.freebsd.org>
next in thread | raw e-mail | index | archive | help
Author: delphij Date: Fri Jan 23 17:41:34 2015 New Revision: 277576 URL: https://svnweb.freebsd.org/changeset/base/277576 Log: MFC r275738: MFV r275546: Reduce scrub activities when system there is enough dirty data, namely when dirty data is more than zfs_vdev_async_write_active_min_dirty_percent (once we start to increase the number of concurrent async writes). While there also correct rounding error which would make scrub end up pausing for (zfs_txg_timeout + 1) seconds instead of the desired zfs_txg_timeout seconds. Illumos issue: 5351 scrub goes for an extra second each txg 5352 scrub should pause when there is some dirty data Modified: stable/10/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_scan.c Directory Properties: stable/10/ (props changed) Modified: stable/10/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_scan.c ============================================================================== --- stable/10/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_scan.c Fri Jan 23 17:31:41 2015 (r277575) +++ stable/10/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_scan.c Fri Jan 23 17:41:34 2015 (r277576) @@ -423,12 +423,11 @@ dsl_scan_sync_state(dsl_scan_t *scn, dmu &scn->scn_phys, tx)); } +extern int zfs_vdev_async_write_active_min_dirty_percent; + static boolean_t dsl_scan_check_pause(dsl_scan_t *scn, const zbookmark_phys_t *zb) { - uint64_t elapsed_nanosecs; - unsigned int mintime; - /* we never skip user/group accounting objects */ if (zb && (int64_t)zb->zb_object < 0) return (B_FALSE); @@ -443,12 +442,28 @@ dsl_scan_check_pause(dsl_scan_t *scn, co if (zb && zb->zb_level != 0) return (B_FALSE); - mintime = (scn->scn_phys.scn_func == POOL_SCAN_RESILVER) ? + /* + * We pause if: + * - we have scanned for the maximum time: an entire txg + * timeout (default 5 sec) + * or + * - we have scanned for at least the minimum time (default 1 sec + * for scrub, 3 sec for resilver), and either we have sufficient + * dirty data that we are starting to write more quickly + * (default 30%), or someone is explicitly waiting for this txg + * to complete. + * or + * - the spa is shutting down because this pool is being exported + * or the machine is rebooting. + */ + int mintime = (scn->scn_phys.scn_func == POOL_SCAN_RESILVER) ? zfs_resilver_min_time_ms : zfs_scan_min_time_ms; - elapsed_nanosecs = gethrtime() - scn->scn_sync_start_time; - if (elapsed_nanosecs / NANOSEC > zfs_txg_timeout || + uint64_t elapsed_nanosecs = gethrtime() - scn->scn_sync_start_time; + int dirty_pct = scn->scn_dp->dp_dirty_total * 100 / zfs_dirty_data_max; + if (elapsed_nanosecs / NANOSEC >= zfs_txg_timeout || (NSEC2MSEC(elapsed_nanosecs) > mintime && - txg_sync_waiting(scn->scn_dp)) || + (txg_sync_waiting(scn->scn_dp) || + dirty_pct >= zfs_vdev_async_write_active_min_dirty_percent)) || spa_shutting_down(scn->scn_dp->dp_spa)) { if (zb) { dprintf("pausing at bookmark %llx/%llx/%llx/%llx\n",
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201501231741.t0NHfY7u097857>