aboutsummaryrefslogtreecommitdiffstats
path: root/module/zfs/metaslab.c
Commit message (Collapse)AuthorAgeFilesLines
* OpenZFS 7090 - zfs should throttle allocationsDon Brady2016-10-131-51/+289
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | OpenZFS 7090 - zfs should throttle allocations Authored by: George Wilson <[email protected]> Reviewed by: Alex Reece <[email protected]> Reviewed by: Christopher Siden <[email protected]> Reviewed by: Dan Kimmel <[email protected]> Reviewed by: Matthew Ahrens <[email protected]> Reviewed by: Paul Dagnelie <[email protected]> Reviewed by: Prakash Surya <[email protected]> Reviewed by: Sebastien Roy <[email protected]> Approved by: Matthew Ahrens <[email protected]> Ported-by: Don Brady <[email protected]> Reviewed-by: Brian Behlendorf <[email protected]> When write I/Os are issued, they are issued in block order but the ZIO pipeline will drive them asynchronously through the allocation stage which can result in blocks being allocated out-of-order. It would be nice to preserve as much of the logical order as possible. In addition, the allocations are equally scattered across all top-level VDEVs but not all top-level VDEVs are created equally. The pipeline should be able to detect devices that are more capable of handling allocations and should allocate more blocks to those devices. This allows for dynamic allocation distribution when devices are imbalanced as fuller devices will tend to be slower than empty devices. The change includes a new pool-wide allocation queue which would throttle and order allocations in the ZIO pipeline. The queue would be ordered by issued time and offset and would provide an initial amount of allocation of work to each top-level vdev. The allocation logic utilizes a reservation system to reserve allocations that will be performed by the allocator. Once an allocation is successfully completed it's scheduled on a given top-level vdev. Each top-level vdev maintains a maximum number of allocations that it can handle (mg_alloc_queue_depth). The pool-wide reserved allocations (top-levels * mg_alloc_queue_depth) are distributed across the top-level vdevs metaslab groups and round robin across all eligible metaslab groups to distribute the work. As top-levels complete their work, they receive additional work from the pool-wide allocation queue until the allocation queue is emptied. OpenZFS-issue: https://www.illumos.org/issues/7090 OpenZFS-commit: https://github.com/openzfs/openzfs/commit/4756c3d7 Closes #5258 Porting Notes: - Maintained minimal stack in zio_done - Preserve linux-specific io sizes in zio_write_compress - Added module params and documentation - Updated to use optimize AVL cmp macros
* Performance optimization of AVL tree comparator functionsGvozden Neskovic2016-08-311-27/+11
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | perf: 2.75x faster ddt_entry_compare() First 256bits of ddt_key_t is a block checksum, which are expected to be close to random data. Hence, on average, comparison only needs to look at first few bytes of the keys. To reduce number of conditional jump instructions, the result is computed as: sign(memcmp(k1, k2)). Sign of an integer 'a' can be obtained as: `(0 < a) - (a < 0)` := {-1, 0, 1} , which is computed efficiently. Synthetic performance evaluation of original and new algorithm over 1G random keys on 2.6GHz Intel(R) Xeon(R) CPU E5-2660 v3: old 6.85789 s new 2.49089 s perf: 2.8x faster vdev_queue_offset_compare() and vdev_queue_timestamp_compare() Compute the result directly instead of using conditionals perf: zfs_range_compare() Speedup between 1.1x - 2.5x, depending on compiler version and optimization level. perf: spa_error_entry_compare() `bcmp()` is not suitable for comparator use. Use `memcmp()` instead. perf: 2.8x faster metaslab_compare() and metaslab_rangesize_compare() perf: 2.8x faster zil_bp_compare() perf: 2.8x faster mze_compare() perf: faster dbuf_compare() perf: faster compares in spa_misc perf: 2.8x faster layout_hash_compare() perf: 2.8x faster space_reftree_compare() perf: libzfs: faster avl tree comparators perf: guid_compare() perf: dsl_deadlist_compare() perf: perm_set_compare() perf: 2x faster range_tree_seg_compare() perf: faster unique_compare() perf: faster vdev_cache _compare() perf: faster vdev_uberblock_compare() perf: faster fuid _compare() perf: faster zfs_znode_hold_compare() Signed-off-by: Gvozden Neskovic <[email protected]> Signed-off-by: Richard Elling <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #5033
* Fix gcc self-comparison warningBrian Behlendorf2016-08-021-1/+1
| | | | | | | | | | | | | | As of gcc 6.1.1 20160621 (Red Hat 6.1.1-3) a self-comparison is detected by gcc in metaslab_alloc(). Resolve the warning by passing a physical size of 0 to BP_SET_BIRTH() as it done by other callers. module/zfs/metaslab.c: In function ‘metaslab_alloc’: module/zfs/metaslab.c:2575:184: error: self-comparison always evaluates to true [-Werror=tautological-compare] Signed-off-by: Brian Behlendorf <[email protected]> Signed-off-by: Gvozden Neskovic <[email protected]> Issue #4907
* gcc build error: -Wbool-compare in metaslab.cDHE2016-03-301-1/+2
| | | | | | | | | | | When debugging is enabled on a very recent version of gcc (tested with 5.3.0), DVA_SET_GANG(dva, !!(flags)) fails because an assertion causes a comparison between what is technically a boolean and an integer. Signed-off-by: DHE <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #4465
* Remove fastwrite mutexRichard Yao2016-01-151-9/+0
| | | | | | | | | | | | | | | | The fast write mutex is intended to protect accounting, but it is redundant because all accounting is performed through atomic operations. It also serializes all metaslab IO behind a mutex, which introduces a theoretical scaling regression that the Illumos developers did not like when we showed this to them. Removing it makes the selection of the metaslab_group lock free as it is on Illumos. The selection is not quite the same without the lock because the loop races with IO completions, but any imbalances caused by this are likely to be corrected by subsequent metaslab group selections. Signed-off-by: Richard Yao <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #3643
* Illumos 6295 - metaslab_condense's dbgmsg should include vdev idJoe Stein2016-01-121-5/+6
| | | | | | | | | | | | | | | | | 6295 metaslab_condense's dbgmsg should include vdev id Reviewed by: George Wilson <[email protected]> Reviewed by: Matthew Ahrens <[email protected]> Reviewed by: Andriy Gapon <[email protected]> Reviewed by: Xin Li <[email protected]> Reviewed by: Justin Gibbs <[email protected]> Approved by: Richard Lowe <[email protected]> References: https://www.illumos.org/issues/6295 https://github.com/illumos/illumos-gate/commit/daec38e Ported-by: kernelOfTruth [email protected] Signed-off-by: Brian Behlendorf <[email protected]>
* Disable LBA weighting on files and SSDsRichard Yao2015-09-011-1/+1
| | | | | | | | | | | | | | The LBA weighting makes sense on rotational media where the outer tracks have twice the bandwidth of the inner tracks. However, it is detrimental on nonrotational media such as solid state disks, where the only effect is to ensure that metaslabs enter the best-fit allocation behavior sooner, which is detrimental to performance. It also makes no sense on files where the underlying filesystem can arrange things however it wants. Signed-off-by: Richard Yao <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #3712
* Align thread priority with Linux defaultsBrian Behlendorf2015-07-281-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Under Linux filesystem threads responsible for handling I/O are normally created with the maximum priority. Non-I/O filesystem processes run with the default priority. ZFS should adopt the same priority scheme under Linux to maintain good performance and so that it will complete fairly when other Linux filesystems are active. The priorities have been updated to the following: $ ps -eLo rtprio,cls,pid,pri,nice,cmd | egrep 'z_|spl_|zvol|arc|dbu|meta' - TS 10743 19 -20 [spl_kmem_cache] - TS 10744 19 -20 [spl_system_task] - TS 10745 19 -20 [spl_dynamic_tas] - TS 10764 19 0 [dbu_evict] - TS 10765 19 0 [arc_prune] - TS 10766 19 0 [arc_reclaim] - TS 10767 19 0 [arc_user_evicts] - TS 10768 19 0 [l2arc_feed] - TS 10769 39 0 [z_unmount] - TS 10770 39 -20 [zvol] - TS 11011 39 -20 [z_null_iss] - TS 11012 39 -20 [z_null_int] - TS 11013 39 -20 [z_rd_iss] - TS 11014 39 -20 [z_rd_int_0] - TS 11022 38 -19 [z_wr_iss] - TS 11023 39 -20 [z_wr_iss_h] - TS 11024 39 -20 [z_wr_int_0] - TS 11032 39 -20 [z_wr_int_h] - TS 11033 39 -20 [z_fr_iss_0] - TS 11041 39 -20 [z_fr_int] - TS 11042 39 -20 [z_cl_iss] - TS 11043 39 -20 [z_cl_int] - TS 11044 39 -20 [z_ioctl_iss] - TS 11045 39 -20 [z_ioctl_int] - TS 11046 39 -20 [metaslab_group_] - TS 11050 19 0 [z_iput] - TS 11121 38 -19 [z_wr_iss] Note that under Linux the meaning of a processes priority is inverted with respect to illumos. High values on Linux indicate a _low_ priority while high value on illumos indicate a _high_ priority. In order to preserve the logical meaning of the minclsyspri and maxclsyspri macros when they are used by the illumos wrapper functions their values have been inverted. This way when changes are merged from upstream illumos we won't need to remember to invert the macro. It could also lead to confusion. This patch depends on https://github.com/zfsonlinux/spl/pull/466. Signed-off-by: Brian Behlendorf <[email protected]> Signed-off-by: Ned Bass <[email protected]> Closes #3607
* Prevent reclaim in metaslab preload threadsTim Chase2015-07-061-0/+2
| | | | | | | | | Reclaim during metaslab preloading can cause deadlocks involving znode z_lock and ARC buffer header ht_lock. Signed-off-by: Tim Chase <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #3532.
* Update all default taskq settingsBrian Behlendorf2015-06-251-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Over the years the default values for the taskqs used on Linux have differed slightly from illumos. In the vast majority of cases this was done to avoid creating an obnoxious number of idle threads which would pollute the process listing. With the addition of support for dynamic taskqs all multi-threaded queues should be created as dynamic taskqs. This allows us to get the best of both worlds. * The illumos default values for the I/O pipeline can be restored. These values are known to work well for most workloads. The only exception is the zio write interrupt taskq which is changed to ZTI_P(12, 8). At least under Linux more threads has been shown to improve performance, see commit 7e55f4e. * Reduces the number of idle threads on the system when it's not under heavy load. The maximum number of threads will only be created when they are required. * Remove the vdev_file_taskq and rely on the system_taskq instead which is now dynamic and may have up to 64-threads. Again this brings us back inline with upstream. * Tasks dispatched with taskq_dispatch_ent() are allowed to use dynamic taskqs. The Linux taskq implementation supports this. Signed-off-by: Brian Behlendorf <[email protected]> Signed-off-by: Tim Chase <[email protected]> Closes #3507
* Make metaslab_aliquot a module parameter.Etienne Dechamps2015-06-221-1/+4
| | | | | | | | | | This seems generally useful. metaslab_aliquot is the ZFS allocation granularity, which is roughly equivalent to what is called the stripe size in traditional RAID arrays. It seems relevant to performance tuning. Signed-off-by: Etienne Dechamps <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]>
* Document metaslab_aliquot.Etienne Dechamps2015-06-221-0/+7
| | | | | Signed-off-by: Etienne Dechamps <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]>
* Allocate disk space fairly in the presence of vdevs of unequal size.Etienne Dechamps2015-06-221-11/+25
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The metaslab allocator device selection algorithm contains a bias mechanism whose goal is to achieve roughly equal disk space usage across all top-level vdevs. It seems that the initial rationale for this code was to allow newly added (empty) vdevs to "come up to speed" faster in an attempt to make the pool quickly converge to a steady state where all vdevs are equally utilized. While the code seems to work reasonably well for this use case, there is another scenario in which this algorithm fails miserably: the case where top-level vdevs don't have the same sizes (capacities). ZFS allows this, and it is a good feature to have, so that users who simply want to build a pool with the disks they happen to have lying around can do so even if the disks have heteregenous sizes. Here's a script that simulates a pool with two vdevs, with one 4X larger than the other: dd if=/dev/zero of=/tmp/d1 bs=1 count=1 seek=134217728 dd if=/dev/zero of=/tmp/d2 bs=1 count=1 seek=536870912 zpool create testspace /tmp/d1 /tmp/d2 dd if=/dev/zero of=/testspace/foobar bs=1M count=256 zpool iostat -v testspace Before this commit, the script would output the following: capacity pool alloc free ---------- ----- ----- testspace 252M 375M /tmp/d1 104M 18.5M /tmp/d2 148M 356M ---------- ----- ----- This demonstrates that the current code handles this situation very poorly: d1 shows 85% usage despite the pool itself being only 40% full. d1 is quite saturated at this point, and is slowing down the entire pool due to saturation, fragmentation and the like. In contrast, here's the result with the code in this commit: capacity pool alloc free ---------- ----- ----- testspace 252M 375M /tmp/d1 56.7M 66.3M /tmp/d2 195M 309M ---------- ----- ------ This looks much better. d1 is 46% used, which is close to the overall pool utilization (40%). The code still doesn't result in perfectly balanced allocation, probably because of the way mg_bias is applied which does not guarantee perfect accuracy, but this is still much better than before. Signed-off-by: Etienne Dechamps <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #3389
* Use taskq_wait_outstanding() functionBrian Behlendorf2015-06-111-2/+2
| | | | | | | | | | | Replace taskq_wait() with taskq_wait_oustanding(). This way callers will only block until previously submitted tasks have been completed. This was the previous behavior of task_wait() prior to the introduction of taskq_wait_outstanding() so this isn't really a functionalty change for these callers. Signed-off-by: Tim Chase <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]>
* Remove metaslab_min_alloc_size module optionBrian Behlendorf2015-05-111-6/+0
| | | | | | | | The metaslab_min_alloc_size option is no longer used in the code. This functionality was removed by commit f3a7f66 and the module options should have been dropped at that time. Signed-off-by: Brian Behlendorf <[email protected]>
* Skip bad DVAs during free by setting zfs_recover=1Brian Behlendorf2015-02-131-6/+4
| | | | | | | | | | | | | | | | | | | | | | When a bad DVA is encountered in metaslab_free_dva() the system should treat it as fatal. This indicates that somehow a damaged DVA was written to disk and that should be impossible. However, we have seen a handful of reports over the years of pools somehow being damaged in this way. Since this damage can render otherwise intact pools unimportable, and the consequence of skipping the bad DVA is only leaked free space, it makes sense to provide a mechanism to ignore the bad DVA. Setting the zfs_recover=1 module option will cause the DVA to be ignored which may allow the pool to be imported. Since zfs_recover=0 by default any pool attempting to free a bad DVA will treat it as a fatal error preserving the current behavior. Signed-off-by: Brian Behlendorf <[email protected]> Closes #3099 Issue #3090 Issue #2720
* Revert "Don't read space maps during import for readonly pools"Brian Behlendorf2015-02-091-3/+2
| | | | | | | | | | This reverts commit 7fc8c33ede10f7104ca0e91d690d3ebb5236887b which accidentally introduced a ztest failure. ztest: '/usr/sbin/zdb -bcc -d -U /var/tmp/zpool.cache ztest' exit code 2 child exited with code 3 Signed-off-by: Brian Behlendorf <[email protected]>
* Don't read space maps during import for readonly poolsBrian Behlendorf2015-02-091-2/+3
| | | | | | | | | | | | | | | | | | Normally when importing a pool the space maps for all top level vdevs are read from disk. The space maps will be required latter when an allocation is performed and free blocks need to be located. However, if the pool is imported readonly then we are guaranteed that no allocations can occur. In this case the space maps need not be loaded.. A similar argument can be made for the DTLs (dirty time logs). Because a pool import will fail if the space maps cannot be read. The ability to safely ignore them makes it more likely that a damaged pool can be imported readonly to recover its contents. Signed-off-by: Brian Behlendorf <[email protected]> Issue #2831
* Change KM_PUSHPAGE -> KM_SLEEPBrian Behlendorf2015-01-161-6/+6
| | | | | | | | | | | | | | | By marking DMU transaction processing contexts with PF_FSTRANS we can revert the KM_PUSHPAGE -> KM_SLEEP changes. This brings us back in line with upstream. In some cases this means simply swapping the flags back. For others fnvlist_alloc() was replaced by nvlist_alloc(..., KM_PUSHPAGE) and must be reverted back to fnvlist_alloc() which assumes KM_SLEEP. The one place KM_PUSHPAGE is kept is when allocating ARC buffers which allows us to dip in to reserved memory. This is again the same as upstream. Signed-off-by: Brian Behlendorf <[email protected]>
* Illumos 5213 - panic in metaslab_init due to space_map_open returning ENXIOPrakash Surya2014-11-141-23/+33
| | | | | | | | | | | | | | | | | | 5213 panic in metaslab_init due to space_map_open returning ENXIO Reviewed by: Matthew Ahrens [email protected] Reviewed by: George Wilson [email protected] References: https://www.illumos.org/issues/5213 https://reviews.csiden.org/r/110 Porting notes: For the Linux port, KM_SLEEP was replaced with KM_PUSHPAGE. Ported by: Turbo Fredriksson <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #2745
* Illumos 5164-5165 - space map fixesMatthew Ahrens2014-10-231-4/+10
| | | | | | | | | | | | | | | | | | | | | | | | | | | | 5164 space_map_max_blksz causes panic, does not work 5165 zdb fails assertion when run on pool with recently-enabled space map_histogram feature Reviewed by: Christopher Siden <[email protected]> Reviewed by: George Wilson <[email protected]> Reviewed by: Saso Kiselkov <[email protected]> Approved by: Dan McDonald <[email protected]> References: https://www.illumos.org/issues/5164 https://www.illumos.org/issues/5165 https://github.com/illumos/illumos-gate/commit/b1be289 Porting Notes: The metaslab_fragmentation() hunk was dropped from this patch because it was already resolved by commit 8b0a084. The comment modified in metaslab.c was updated to use the correct variable name, space_map_blksz. The upstream commit incorrectly used space_map_blksize. Ported by: Turbo Fredriksson <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #2697
* Illumos 4958 zdb trips assert on pools with ashift >= 0xeAlex Reece2014-10-231-4/+35
| | | | | | | | | | | | | | | | | | | | | | 4958 zdb trips assert on pools with ashift >= 0xe Reviewed by: Matthew Ahrens <[email protected]> Reviewed by: Max Grossman <[email protected]> Reviewed by: George Wilson <[email protected]> Reviewed by: Christopher Siden <[email protected]> Approved by: Garrett D'Amore <[email protected]> References: https://www.illumos.org/issues/4958 https://github.com/illumos/illumos-gate/commit/2a104a5 Porting notes: Keep the ZIO_FLAG_FASTWRITE define. This is for a feature present in Linux but not yet in *BSD. Ported by: Turbo Fredriksson <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #2697
* Don't upgrade a metaslab when the pool is not writableTim Chase2014-08-181-5/+8
| | | | | | | | | | | | | | Illumos 4982 added code to metaslab_fragmentation() to proactively update space maps when the spacemap_histogram feature is enabled. This should only happen when the pool is writeable. References: https://www.illumos.org/issues/4982 https://github.com/illumos/illumos-gate/commit/2e4c998 Signed-off-by: Tim Chase <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #2595
* Illumos 4976-4984 - metaslab improvementsGeorge Wilson2014-08-181-135/+496
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 4976 zfs should only avoid writing to a failing non-redundant top-level vdev 4978 ztest fails in get_metaslab_refcount() 4979 extend free space histogram to device and pool 4980 metaslabs should have a fragmentation metric 4981 remove fragmented ops vector from block allocator 4982 space_map object should proactively upgrade when feature is enabled 4983 need to collect metaslab information via mdb 4984 device selection should use fragmentation metric Reviewed by: Matthew Ahrens <[email protected]> Reviewed by: Adam Leventhal <[email protected]> Reviewed by: Christopher Siden <[email protected]> Approved by: Garrett D'Amore <[email protected]> References: https://www.illumos.org/issues/4976 https://www.illumos.org/issues/4978 https://www.illumos.org/issues/4979 https://www.illumos.org/issues/4980 https://www.illumos.org/issues/4981 https://www.illumos.org/issues/4982 https://www.illumos.org/issues/4983 https://www.illumos.org/issues/4984 https://github.com/illumos/illumos-gate/commit/2e4c998 Notes: The "zdb -M" option has been re-tasked to display the new metaslab fragmentation metric and the new "zdb -I" option is used to control the maximum number of in-flight I/Os. The new fragmentation metric is derived from the space map histogram which has been rolled up to the vdev and pool level and is presented to the user via "zpool list". Add a number of module parameters related to the new metaslab weighting logic. Ported by: Tim Chase <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #2595
* Illumos 4757, 4913Matthew Ahrens2014-08-011-0/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | 4757 ZFS embedded-data block pointers ("zero block compression") 4913 zfs release should not be subject to space checks Reviewed by: Adam Leventhal <[email protected]> Reviewed by: Max Grossman <[email protected]> Reviewed by: George Wilson <[email protected]> Reviewed by: Christopher Siden <[email protected]> Reviewed by: Dan McDonald <[email protected]> Approved by: Dan McDonald <[email protected]> References: https://www.illumos.org/issues/4757 https://www.illumos.org/issues/4913 https://github.com/illumos/illumos-gate/commit/5d7b4d4 Porting notes: For compatibility with the fastpath code the zio_done() function needed to be updated. Because embedded-data block pointers do not require DVAs to be allocated the associated vdevs will not be marked and therefore should not be unmarked. Ported by: Tim Chase <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #2544
* Illumos 4754, 4755George Wilson2014-07-301-41/+6
| | | | | | | | | | | | | | | | | | | 4754 io issued to near-full luns even after setting noalloc threshold 4755 mg_alloc_failures is no longer needed Reviewed by: Matthew Ahrens <[email protected]> Reviewed by: Adam Leventhal <[email protected]> Reviewed by: Dan McDonald <[email protected]> Approved by: Dan McDonald <[email protected]> References: https://www.illumos.org/issues/4754 https://www.illumos.org/issues/4755 https://github.com/illumos/illumos-gate/commit/b6240e8 Ported by: Tim Chase <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #2533
* Illumos #4374Matthew Ahrens2014-07-301-5/+5
| | | | | | | | | | | | | | | | | | | 4374 dn_free_ranges should use range_tree_t Reviewed by: George Wilson <[email protected]> Reviewed by: Max Grossman <[email protected]> Reviewed by: Christopher Siden <[email protected] Reviewed by: Garrett D'Amore <[email protected]> Reviewed by: Dan McDonald <[email protected]> Approved by: Dan McDonald <[email protected]> References: https://www.illumos.org/issues/4374 https://github.com/illumos/illumos-gate/commit/bf16b11 Ported by: Tim Chase <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #2531
* Illumos #4756 Fix metaslab_group_preload deadlockGeorge Wilson2014-07-221-3/+22
| | | | | | | | | | | | | | | | | | | | | | | 4756 metaslab_group_preload() could deadlock Reviewed by: Matthew Ahrens <[email protected]> Reviewed by: Christopher Siden <[email protected]> Reviewed by: Dan McDonald <[email protected]> Reviewed by: Saso Kiselkov <[email protected]> Approved by: Garrett D'Amore <[email protected]> The metaslab_group_preload() function grabs the mg_lock and then later tries to grab the metaslab lock. This lock ordering may lead to a deadlock since other consumers of the mg_lock will grab the metaslab lock first. References: https://www.illumos.org/issues/4756 https://github.com/illumos/illumos-gate/commit/30beaff Ported-by: Prakash Surya <[email protected]> Signed-off-by: Prakash Surya <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #2488
* Illumos #4730 destroy metaslab group taskqGeorge Wilson2014-07-221-1/+2
| | | | | | | | | | | | | | | | | | | | | | | | | 4730 metaslab group taskq should be destroyed in metaslab_group_destroy() Reviewed by: Alex Reece <[email protected]> Reviewed by: Matthew Ahrens <[email protected]> Reviewed by: Sebastien Roy <[email protected]> Reviewed by: Rich Lowe <[email protected]> Reviewed by: Dan McDonald <[email protected]> Approved by: Dan McDonald <[email protected]> References: https://www.illumos.org/issues/4730 https://github.com/illumos/illumos-gate/commit/be08211 Porting notes: Under ZFSonlinux, one of the effects of not destroying the taskq is that zdb would never exit (due to the SPL taskq implementation). Ported-by: Tim Chase <[email protected]> Signed-off-by: Prakash Surya <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #2488
* Illumos #4101, #4102, #4103, #4105, #4106George Wilson2014-07-221-531/+733
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 4101 metaslab_debug should allow for fine-grained control 4102 space_maps should store more information about themselves 4103 space map object blocksize should be increased 4105 removing a mirrored log device results in a leaked object 4106 asynchronously load metaslab Reviewed by: Matthew Ahrens <[email protected]> Reviewed by: Adam Leventhal <[email protected]> Reviewed by: Sebastien Roy <[email protected]> Approved by: Garrett D'Amore <[email protected]> Prior to this patch, space_maps were preferred solely based on the amount of free space left in each. Unfortunately, this heuristic didn't contain any information about the make-up of that free space, which meant we could keep preferring and loading a highly fragmented space map that wouldn't actually have enough contiguous space to satisfy the allocation; then unloading that space_map and repeating the process. This change modifies the space_map's to store additional information about the contiguous space in the space_map, so that we can use this information to make a better decision about which space_map to load. This requires reallocating all space_map objects to increase their bonus buffer size sizes enough to fit the new metadata. The above feature can be enabled via a new feature flag introduced by this change: com.delphix:spacemap_histogram In addition to the above, this patch allows the space_map block size to be increase. Currently the block size is set to be 4K in size, which has certain implications including the following: * 4K sector devices will not see any compression benefit * large space_maps require more metadata on-disk * large space_maps require more time to load (typically random reads) Now the space_map block size can adjust as needed up to the maximum size set via the space_map_max_blksz variable. A bug was fixed which resulted in potentially leaking an object when removing a mirrored log device. The previous logic for vdev_remove() did not deal with removing top-level vdevs that are interior vdevs (i.e. mirror) correctly. The problem would occur when removing a mirrored log device, and result in the DTL space map object being leaked; because top-level vdevs don't have DTL space map objects associated with them. References: https://www.illumos.org/issues/4101 https://www.illumos.org/issues/4102 https://www.illumos.org/issues/4103 https://www.illumos.org/issues/4105 https://www.illumos.org/issues/4106 https://github.com/illumos/illumos-gate/commit/0713e23 Porting notes: A handful of kmem_alloc() calls were converted to kmem_zalloc(). Also, the KM_PUSHPAGE and TQ_PUSHPAGE flags were used as necessary. Ported-by: Tim Chase <[email protected]> Signed-off-by: Prakash Surya <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #2488
* Move metaslab_group_alloc_update() callPrakash Surya2014-07-221-2/+2
| | | | | | | | | | | | This changes moves the called to metaslab_group_alloc_update() to the metaslab_sync_reassess() function. The original placement of the call within metaslab_sync_done() appears to have been a simple mistake, introduced by ac72fac3eaa569902cad88053167f7d74e7fe7e4. This aligns us more closely to the upstream illumos code base. Signed-off-by: Prakash Surya <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]>
* Convert zfs_mg_noalloc_threshold to a module parameter and documentTim Chase2014-07-161-0/+4
| | | | | | | | | | The parameter was added as illumos issue 4081 which was committed to zfsonlinux in ac72fac3eaa569902cad88053167f7d74e7fe7e4. This patch documents the parameter and allows for it to be set as a module parameter. Signed-off-by: Tim Chase <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #2483
* Fix styleBrian Behlendorf2014-05-061-1/+2
| | | | | | | A minor style issue was accidentally introduced by aa7d06a. This change resolves that style problem. Signed-off-by: Brian Behlendorf <[email protected]>
* Illumos #4101 finer-grained control of metaslab_debugGeorge Wilson2014-05-061-6/+14
| | | | | | | | | | | | | | | | | | | | | | | | | | Today the metaslab_debug logic performs two tasks: - load all metaslabs on import/open - don't unload metaslabs at the end of spa_sync This change provides knobs for each of these independently. References: https://illumos.org/issues/4101 https://github.com/illumos/illumos-gate/commit/0713e23 Notes: 1) This is a small piece of the metaslab improvement patch from Illumos. It was worth bringing over before the rest, since it's low risk and it can be useful on fragmented pools (e.g. Lustre MDTs). metaslab_debug_unload would give the performance benefit of the old metaslab_debug option without causing unwanted delay during pool import. Ported-by: Ned Bass <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #2227
* cstyle: Resolve C style issuesMichael Kjorling2013-12-181-3/+5
| | | | | | | | | | | | | | | | | | The vast majority of these changes are in Linux specific code. They are the result of not having an automated style checker to validate the code when it was originally written. Others were caused when the common code was slightly adjusted for Linux. This patch contains no functional changes. It only refreshes the code to conform to style guide. Everyone submitting patches for inclusion upstream should now run 'make checkstyle' and resolve any warning prior to opening a pull request. The automated builders have been updated to fail a build if when 'make checkstyle' detects an issue. Signed-off-by: Brian Behlendorf <[email protected]> Closes #1821
* Illumos #4168, #4169, #4170George Wilson2013-11-051-1/+9
| | | | | | | | | | | | | | | | | | | | 4168 ztest assertion failure in dbuf_undirty 4169 verbatim import causes zdb to segfault 4170 zhack leaves pool in ACTIVE state Reviewed by: Adam Leventhal <[email protected]> Reviewed by: Eric Schrock <[email protected]> Reviewed by: Matthew Ahrens <[email protected]> Approved by: Dan McDonald <[email protected]> References: https://www.illumos.org/issues/4168 https://www.illumos.org/issues/4169 https://www.illumos.org/issues/4170 illumos/illumos-gate@7fdd916c474ea52896c671bbe7b56ba34a1ca132 Ported-by: Richard Yao <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Issue #1775
* Illumos #3954, #4080, #4081George Wilson2013-11-051-4/+109
| | | | | | | | | | | | | | | | | | | 3954 metaslabs continue to load even after hitting zfs_mg_alloc_failure limit 4080 zpool clear fails to clear pool 4081 need zfs_mg_noalloc_threshold Reviewed by: Adam Leventhal <[email protected]> Reviewed by: Matthew Ahrens <[email protected]> Approved by: Richard Lowe <[email protected]> References: https://www.illumos.org/issues/3954 https://www.illumos.org/issues/4080 https://www.illumos.org/issues/4081 illumos/illumos-gate@22e30981d82a0b6dc89253596ededafae8655e00 Ported-by: Richard Yao <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Issue #1775
* Illumos #3598Matthew Ahrens2013-10-311-6/+8
| | | | | | | | | | | | | | | | | | | | | | | | | | | 3598 want to dtrace when errors are generated in zfs Reviewed by: Dan Kimmel <[email protected]> Reviewed by: Adam Leventhal <[email protected]> Reviewed by: Christopher Siden <[email protected]> Approved by: Garrett D'Amore <[email protected]> References: https://www.illumos.org/issues/3598 illumos/illumos-gate@be6fd75a69ae679453d9cda5bff3326111e6d1ca Ported-by: Richard Yao <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Issue #1775 Porting notes: 1. include/sys/zfs_context.h has been modified to render some new macros inert until dtrace is available on Linux. 2. Linux-specific changes have been adapted to use SET_ERROR(). 3. I'm NOT happy about this change. It does nothing but ugly up the code under Linux. Unfortunately we need to take it to avoid more merge conflicts in the future. -Brian
* Illumos #3578, #3579George Wilson2013-10-311-11/+27
| | | | | | | | | | | | | | | | | | | 3578 transferring the freed map to the defer map should be constant time 3579 ztest trips assertion in metaslab_weight() Reviewed by: Matthew Ahrens <[email protected]> Reviewed by: Dan Kimmel <[email protected]> Reviewed by: Adam Leventhal <[email protected]> Reviewed by: Christopher Siden <[email protected]> Reviewed by: Richard Elling <[email protected]> Approved by: Dan McDonald <[email protected]> References: https://www.illumos.org/issues/3578 https://www.illumos.org/issues/3579 illumos/illumos-gate@9eb57f7f3fbb970d4b9b89dcd5ecf543fe2414d5 Ported-by: Richard Yao <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]>
* Illumos #3552George Wilson2013-10-181-10/+17
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 3552 condensing one space map burns 3 seconds of CPU in spa_sync() thread (fix race condition) References: https://www.illumos.org/issues/3552 illumos/illumos-gate@03f8c366886542ed249a15d755ae78ea4e775d9d Ported-by: Richard Yao <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Porting notes: This fixes an upstream regression that was introduced in commit zfsonlinux/zfs@e51be06697762215dc3b679f8668987034a5a048, which ported the Illumos 3552 changes. This fix was added to upstream rather quickly, but at the time of the port, no one spotted it and the race was rare enough that it passed our regression tests. I discovered this when comparing our metaslab.c to the illumos metaslab.c. Without this change it is possible for metaslab_group_alloc() to consume a large amount of cpu time. Since this occurs under a mutex in a rcu critical section the kernel will log this to the console as a self-detected cpu stall as follows: INFO: rcu_sched self-detected stall on CPU { 0} (t=60000 jiffies g=11431890 c=11431889 q=18271) Closes #1687 Closes #1720 Closes #1731 Closes #1747
* Illumos #3464Matthew Ahrens2013-09-041-0/+40
| | | | | | | | | | | | | | | | | 3464 zfs synctask code needs restructuring Reviewed by: Dan Kimmel <[email protected]> Reviewed by: Adam Leventhal <[email protected]> Reviewed by: George Wilson <[email protected]> Reviewed by: Christopher Siden <[email protected]> Approved by: Garrett D'Amore <[email protected]> References: https://www.illumos.org/issues/3464 illumos/illumos-gate@3b2aab18808792cbd248a12f1edf139b89833c13 Ported-by: Tim Chase <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #1495
* Illumos #3552, #3564George Wilson2013-06-191-85/+265
| | | | | | | | | | | | | | | | | | 3552 condensing one space map burns 3 seconds of CPU in spa_sync() thread 3564 spa_sync() spends 5-10% of its time in metaslab_sync() (when not condensing) Reviewed by: Adam Leventhal <[email protected]> Reviewed by: Dan Kimmel <[email protected]> Reviewed by: Matthew Ahrens <[email protected]> Approved by: Richard Lowe <[email protected]> References: illumos/illumos-gate@16a4a8074274d2d7cc408589cf6359f4a378c861 https://www.illumos.org/issues/3552 https://www.illumos.org/issues/3564 Ported-by: Tim Chase <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #1513
* Illumos #3006Madhav Suresh2013-06-191-1/+1
| | | | | | | | | | | | | | | | | | | | 3006 VERIFY[S,U,P] and ASSERT[S,U,P] frequently check if first argument is zero Reviewed by Matt Ahrens <[email protected]> Reviewed by George Wilson <[email protected]> Approved by Eric Schrock <[email protected]> References: illumos/illumos-gate@fb09f5aad449c97fe309678f3f604982b563a96f https://illumos.org/issues/3006 Requires: zfsonlinux/spl@1c6d149feb4033e4a56fb987004edc5d45288bcb Ported-by: Tim Chase <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Closes #1509
* Illumos #3329, #3330, #3331, #3335George Wilson2013-05-061-2/+3
| | | | | | | | | | | | | | | | | | | | | | | | 3329 spa_sync() spends 10-20% of its time in spa_free_sync_cb() 3330 space_seg_t should have its own kmem_cache 3331 deferred frees should happen after sync_pass 1 3335 make SYNC_PASS_* constants tunable Reviewed by: Adam Leventhal <[email protected]> Reviewed by: Matt Ahrens <[email protected]> Reviewed by: Christopher Siden <[email protected]> Reviewed by: Eric Schrock <[email protected]> Reviewed by: Richard Lowe <[email protected]> Reviewed by: Dan McDonald <[email protected]> Approved by: Eric Schrock <[email protected]> References: illumos/illumos-gate@01f55e48fb4d524eaf70687728aa51b7762e2e97 https://www.illumos.org/issues/3329 https://www.illumos.org/issues/3330 https://www.illumos.org/issues/3331 https://www.illumos.org/issues/3335 Ported-by: Brian Behlendorf <[email protected]>
* Add metaslab_debug optionBrian Behlendorf2013-03-181-1/+6
| | | | | | | | | Enabling metaslab debugging will prevent space maps from being automatically unloaded. This can significantly increase the memory footprint but being able to dynamically control this is helpful for debugging and certain performance testing. Signed-off-by: Brian Behlendorf <[email protected]>
* Add FASTWRITE algorithm for synchronous writes.Etienne Dechamps2012-10-171-2/+70
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Currently, ZIL blocks are spread over vdevs using hint block pointers managed by the ZIL commit code and passed to metaslab_alloc(). Spreading log blocks accross vdevs is important for performance: indeed, using mutliple disks in parallel decreases the ZIL commit latency, which is the main performance metric for synchronous writes. However, the current implementation suffers from the following issues: 1) It would be best if the ZIL module was not aware of such low-level details. They should be handled by the ZIO and metaslab modules; 2) Because the hint block pointer is managed per log, simultaneous commits from multiple logs might use the same vdevs at the same time, which is inefficient; 3) Because dmu_write() does not honor the block pointer hint, indirect writes are not spread. The naive solution of rotating the metaslab rotor each time a block is allocated for the ZIL or dmu_sync() doesn't work in practice because the first ZIL block to be written is actually allocated during the previous commit. Consequently, when metaslab_alloc() decides the vdev for this block, it will do so while a bunch of other allocations are happening at the same time (from dmu_sync() and other ZILs). This means the vdev for this block is chosen more or less at random. When the next commit happens, there is a high chance (especially when the number of blocks per commit is slightly less than the number of the disks) that one disk will have to write two blocks (with a potential seek) while other disks are sitting idle, which defeats spreading and increases the commit latency. This commit introduces a new concept in the metaslab allocator: fastwrites. Basically, each top-level vdev maintains a counter indicating the number of synchronous writes (from dmu_sync() and the ZIL) which have been allocated but not yet completed. When the metaslab is called with the FASTWRITE flag, it will choose the vdev with the least amount of pending synchronous writes. If there are multiple vdevs with the same value, the first matching vdev (starting from the rotor) is used. Once metaslab_alloc() has decided which vdev the block is allocated to, it updates the fastwrite counter for this vdev. The rationale goes like this: when an allocation is done with FASTWRITE, it "reserves" the vdev until the data is written. Until then, all future allocations will naturally avoid this vdev, even after a full rotation of the rotor. As a result, pending synchronous writes at a given point in time will be nicely spread over all vdevs. This contrasts with the previous algorithm, which is based on the implicit assumption that blocks are written instantaneously after they're allocated. metaslab_fastwrite_mark() and metaslab_fastwrite_unmark() are used to manually increase or decrease fastwrite counters, respectively. They should be used with caution, as there is no per-BP tracking of fastwrite information, so leaks and "double-unmarks" are possible. There is, however, an assert in the vdev teardown code which will fire if the fastwrite counters are not zero when the pool is exported or the vdev removed. Note that as stated above, marking is also done implictly by metaslab_alloc(). ZIO also got a new FASTWRITE flag; when it is used, ZIO will pass it to the metaslab when allocating (assuming ZIO does the allocation, which is only true in the case of dmu_sync). This flag will also trigger an unmark when zio_done() fires. A side-effect of the new algorithm is that when a ZIL stops being used, its last block can stay in the pending state (allocated but not yet written) for a long time, polluting the fastwrite counters. To avoid that, I've implemented a somewhat crude but working solution which unmarks these pending blocks in zil_sync(), thus guaranteeing that linguering fastwrites will get pruned at each sync event. The best performance improvements are observed with pools using a large number of top-level vdevs and heavy synchronous write workflows (especially indirect writes and concurrent writes from multiple ZILs). Real-life testing shows a 200% to 300% performance increase with indirect writes and various commit sizes. Signed-off-by: Brian Behlendorf <[email protected]> Issue #1013
* Switch KM_SLEEP to KM_PUSHPAGERichard Yao2012-08-271-5/+5
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Differences between how paging is done on Solaris and Linux can cause deadlocks if KM_SLEEP is used in any the following contexts. * The txg_sync thread * The zvol write/discard threads * The zpl_putpage() VFS callback This is because KM_SLEEP will allow for direct reclaim which may result in the VM calling back in to the filesystem or block layer to write out pages. If a lock is held over this operation the potential exists to deadlock the system. To ensure forward progress all memory allocations in these contexts must us KM_PUSHPAGE which disables performing any I/O to accomplish the memory allocation. Previously, this behavior was acheived by setting PF_MEMALLOC on the thread. However, that resulted in unexpected side effects such as the exhaustion of pages in ZONE_DMA. This approach touchs more of the zfs code, but it is more consistent with the right way to handle these cases under Linux. This is patch lays the ground work for being able to safely revert the following commits which used PF_MEMALLOC: 21ade34 Disable direct reclaim for z_wr_* threads cfc9a5c Fix zpl_writepage() deadlock eec8164 Fix ASSERTION(!dsl_pool_sync_context(tx->tx_pool)) Signed-off-by: Richard Yao <[email protected]> Signed-off-by: Brian Behlendorf <[email protected]> Issue #726
* Illumos #1909: disk sync write perf regression when slog is used post oi_148Brian Behlendorf2012-04-191-2/+2
| | | | | | | | | | | | | | | | | | | Reviewed by: Matt Ahrens <[email protected]> Reviewed by: Eric Schrock <[email protected]> Reviewed by: Robert Mustacchi <[email protected]> Reviewed by: Bill Pijewski <[email protected]> Reviewed by: Richard Elling <[email protected]> Reviewed by: Steve Gonczi <[email protected]> Reviewed by: Garrett D'Amore <[email protected]> Reviewed by: Dan McDonald <[email protected]> Reviewed by: Albert Lee <[email protected]> Approved by: Eric Schrock <[email protected]> Refererces to Illumos issue: https://www.illumos.org/issues/1909 Signed-off-by: Brian Behlendorf <[email protected]> Closes #680
* Illumos #1051: zfs should handle imbalanced lunsGeorge Wilson2011-08-011-27/+78
| | | | | | | | | | | | | | | | | | | | | | Today zfs tries to allocate blocks evenly across all devices. This means when devices are imbalanced zfs will use lots of CPU searching for space on devices which tend to be pretty full. It should instead fail quickly on the full LUNs and move onto devices which have more availability. Reviewed by: Eric Schrock <[email protected]> Reviewed by: Matt Ahrens <[email protected]> Reviewed by: Adam Leventhal <[email protected]> Reviewed by: Albert Lee <[email protected]> Reviewed by: Gordon Ross <[email protected]> Approved by: Garrett D'Amore <[email protected]> References to Illumos issue and patch: - https://www.illumos.org/issues/510 - https://github.com/illumos/illumos-gate/commit/5ead3ed965 Signed-off-by: Brian Behlendorf <[email protected]> Issue #340
* Fix metaslabBrian Behlendorf2010-08-311-0/+20
| | | | | | | | | | | If your only going to allow one allocator to be used and it is defined at compile time there is no point including the others in the build. This patch could/should be refined for Linux to make the metaslab configurable at run time. That might be a bit tricky however since you would need to quiese all IO. Short of that making it configurable as a module load option would be a reasonable compromise. Signed-off-by: Brian Behlendorf <[email protected]>