On Mon, Dec 18, 2023 at 11:05 AM Kairui Song ryncsn@gmail.com wrote:
Yu Zhao yuzhao@google.com 于2023年12月15日周五 12:56写道：
On Thu, Dec 14, 2023 at 04:51:00PM -0700, Yu Zhao wrote:
On Thu, Dec 14, 2023 at 11:38 AM Kairui Song ryncsn@gmail.com wrote:
Yu Zhao yuzhao@google.com 于2023年12月14日周四 11:09写道：
On Wed, Dec 13, 2023 at 12:59:14AM -0700, Yu Zhao wrote:
On Tue, Dec 12, 2023 at 8:03 PM Kairui Song ryncsn@gmail.com wrote: > > Kairui Song ryncsn@gmail.com 于2023年12月12日周二 14:52写道： > > > > Yu Zhao yuzhao@google.com 于2023年12月12日周二 06:07写道： > > > > > > On Fri, Dec 8, 2023 at 1:24 AM Kairui Song ryncsn@gmail.com wrote: > > > > > > > > Yu Zhao yuzhao@google.com 于2023年12月8日周五 14:14写道： > > > > > > > > > > Unmapped folios accessed through file descriptors can be > > > > > underprotected. Those folios are added to the oldest generation based > > > > > on: > > > > > 1. The fact that they are less costly to reclaim (no need to walk the > > > > > rmap and flush the TLB) and have less impact on performance (don't > > > > > cause major PFs and can be non-blocking if needed again). > > > > > 2. The observation that they are likely to be single-use. E.g., for > > > > > client use cases like Android, its apps parse configuration files > > > > > and store the data in heap (anon); for server use cases like MySQL, > > > > > it reads from InnoDB files and holds the cached data for tables in > > > > > buffer pools (anon). > > > > > > > > > > However, the oldest generation can be very short lived, and if so, it > > > > > doesn't provide the PID controller with enough time to respond to a > > > > > surge of refaults. (Note that the PID controller uses weighted > > > > > refaults and those from evicted generations only take a half of the > > > > > whole weight.) In other words, for a short lived generation, the > > > > > moving average smooths out the spike quickly. > > > > > > > > > > To fix the problem: > > > > > 1. For folios that are already on LRU, if they can be beyond the > > > > > tracking range of tiers, i.e., five accesses through file > > > > > descriptors, move them to the second oldest generation to give them > > > > > more time to age. (Note that tiers are used by the PID controller > > > > > to statistically determine whether folios accessed multiple times > > > > > through file descriptors are worth protecting.) > > > > > 2. When adding unmapped folios to LRU, adjust the placement of them so > > > > > that they are not too close to the tail. The effect of this is > > > > > similar to the above. > > > > > > > > > > On Android, launching 55 apps sequentially: > > > > > Before After Change > > > > > workingset_refault_anon 25641024 25598972 0% > > > > > workingset_refault_file 115016834 106178438 -8% > > > > > > > > Hi Yu, > > > > > > > > Thanks you for your amazing works on MGLRU. > > > > > > > > I believe this is the similar issue I was trying to resolve previously: > > > > https://lwn.net/Articles/945266/ > > > > The idea is to use refault distance to decide if the page should be > > > > place in oldest generation or some other gen, which per my test, > > > > worked very well, and we have been using refault distance for MGLRU in > > > > multiple workloads. > > > > > > > > There are a few issues left in my previous RFC series, like anon pages > > > > in MGLRU shouldn't be considered, I wanted to collect feedback or test > > > > cases, but unfortunately it seems didn't get too much attention > > > > upstream. > > > > > > > > I think both this patch and my previous series are for solving the > > > > file pages underpertected issue, and I did a quick test using this > > > > series, for mongodb test, refault distance seems still a better > > > > solution (I'm not saying these two optimization are mutually exclusive > > > > though, just they do have some conflicts in implementation and solving > > > > similar problem): > > > > > > > > Previous result: > > > > ================================================================== > > > > Execution Results after 905 seconds > > > > ------------------------------------------------------------------ > > > > Executed Time (µs) Rate > > > > STOCK_LEVEL 2542 27121571486.2 0.09 txn/s > > > > ------------------------------------------------------------------ > > > > TOTAL 2542 27121571486.2 0.09 txn/s > > > > > > > > This patch: > > > > ================================================================== > > > > Execution Results after 900 seconds > > > > ------------------------------------------------------------------ > > > > Executed Time (µs) Rate > > > > STOCK_LEVEL 1594 27061522574.4 0.06 txn/s > > > > ------------------------------------------------------------------ > > > > TOTAL 1594 27061522574.4 0.06 txn/s > > > > > > > > Unpatched version is always around ~500. > > > > > > Thanks for the test results! > > > > > > > I think there are a few points here: > > > > - Refault distance make use of page shadow so it can better > > > > distinguish evicted pages of different access pattern (re-access > > > > distance). > > > > - Throttled refault distance can help hold part of workingset when > > > > memory is too small to hold the whole workingset. > > > > > > > > So maybe part of this patch and the bits of previous series can be > > > > combined to work better on this issue, how do you think? > > > > > > I'll try to find some time this week to look at your RFC. It'd be a > > Hi Yu, > > I'm working on V4 of the RFC now, which just update some comments, and > skip anon page re-activation in refault path for mglru which was not > very helpful, only some tiny adjustment. > And I found it easier to test with fio, using following test script: > > #!/bin/bash > swapoff -a > > modprobe brd rd_nr=1 rd_size=16777216 > mkfs.ext4 /dev/ram0 > mount /dev/ram0 /mnt > > mkdir -p /sys/fs/cgroup/benchmark > cd /sys/fs/cgroup/benchmark > > echo 4G > memory.max > echo $$ > cgroup.procs > echo 3 > /proc/sys/vm/drop_caches > > fio -name=mglru --numjobs=12 --directory=/mnt --size=1024m \ > --buffered=1 --ioengine=io_uring --iodepth=128 \ > --iodepth_batch_submit=32 --iodepth_batch_complete=32 \ > --rw=randread --random_distribution=zipf:0.5 --norandommap \ > --time_based --ramp_time=5m --runtime=5m --group_reporting > > zipf:0.5 is used here to simulate a cached read with slight bias > towards certain pages. > Unpatched 6.7-rc4: > Run status group 0 (all jobs): > READ: bw=6548MiB/s (6866MB/s), 6548MiB/s-6548MiB/s > (6866MB/s-6866MB/s), io=1918GiB (2060GB), run=300001-300001msec > > Patched with RFC v4: > Run status group 0 (all jobs): > READ: bw=7270MiB/s (7623MB/s), 7270MiB/s-7270MiB/s > (7623MB/s-7623MB/s), io=2130GiB (2287GB), run=300001-300001msec > > Patched with this series: > Run status group 0 (all jobs): > READ: bw=7098MiB/s (7442MB/s), 7098MiB/s-7098MiB/s > (7442MB/s-7442MB/s), io=2079GiB (2233GB), run=300002-300002msec > > MGLRU off: > Run status group 0 (all jobs): > READ: bw=6525MiB/s (6842MB/s), 6525MiB/s-6525MiB/s > (6842MB/s-6842MB/s), io=1912GiB (2052GB), run=300002-300002msec > > - If I change zipf:0.5 to random: > Unpatched 6.7-rc4: > Patched with this series: > Run status group 0 (all jobs): > READ: bw=5975MiB/s (6265MB/s), 5975MiB/s-5975MiB/s > (6265MB/s-6265MB/s), io=1750GiB (1879GB), run=300002-300002msec > > Patched with RFC v4: > Run status group 0 (all jobs): > READ: bw=5987MiB/s (6278MB/s), 5987MiB/s-5987MiB/s > (6278MB/s-6278MB/s), io=1754GiB (1883GB), run=300001-300001msec > > Patched with this series: > Run status group 0 (all jobs): > READ: bw=5839MiB/s (6123MB/s), 5839MiB/s-5839MiB/s > (6123MB/s-6123MB/s), io=1711GiB (1837GB), run=300001-300001msec > > MGLRU off: > Run status group 0 (all jobs): > READ: bw=5689MiB/s (5965MB/s), 5689MiB/s-5689MiB/s > (5965MB/s-5965MB/s), io=1667GiB (1790GB), run=300003-300003msec > > fio uses ramdisk so LRU accuracy will have smaller impact. The Mongodb > test I provided before uses a SATA SSD so it will have a much higher > impact. I'll provides a script to setup the test case and run it, it's > more complex to setup than fio since involving setting up multiple > replicas and auth and hundreds of GB of test fixtures, I'm currently > occupied by some other tasks but will try best to send them out as > soon as possible.
Thanks! Apparently your RFC did show better IOPS with both access patterns, which was a surprise to me because it had higher refaults and usually higher refautls result in worse performance.
And thanks for providing the refaults I requested for -- your data below confirms what I mentioned above:
For fio: Your RFC This series Change workingset_refault_file 628192729 596790506 -5% IOPS 1862k 1830k -2%
For MongoDB: Your RFC This series Change workingset_refault_anon 10512 35277 +30% workingset_refault_file 22751782 20335355 -11% total 22762294 20370632 -11% TPS 0.09 0.06 -33%
For MongoDB, this series should be a big win (but apparently it's not), especially when using zram, since an anon refault should be a lot cheaper than a file refault.
So, I'm baffled...
One important detail I forgot to mention: based on your data from lru_gen_full, I think there is another difference between our Kconfigs:
Your Kconfig My Kconfig Max possibleLRU_REFS_WIDTH 1 2 2
Hi Yu,
Thanks for the info, my fault, I forgot to update my config as I was testing some other features. Buf after I changed LRU_REFS_WIDTH to 2 by disabling IDLE_PAGE, thing got much worse for MongoDB test:
With LRU_REFS_WIDTH == 2:
This patch:
Execution Results after 919 seconds
Executed Time (µs) RateSTOCK_LEVEL 488 27598136201.9 0.02 txn/s
TOTAL 488 27598136201.9 0.02 txn/s
memcg 86 /system.slice/docker-1c3a90be9f0a072f5719332419550cd0e1455f2cd5863bc2780ca4d3f913ece5.scope node 0 1 948187 0x 0x 0 0 0 0 0 0 0· 1 0 0 0 0 0 0· 2 0 0 0 0 0 0· 3 0 0 0 0 0 0· 0 0 0 0 0 0· 2 948187 0 6051788· 0 0r 0e 0p 11916r 66442e 0p 1 0r 0e 0p 903r 16888e 0p 2 0r 0e 0p 459r 9764e 0p 3 0r 0e 0p 0r 0e 2874p 0 0 0 0 0 0· 3 948187 1353160 6351· 0 0 0 0 0 0 0· 1 0 0 0 0 0 0· 2 0 0 0 0 0 0· 3 0 0 0 0 0 0· 0 0 0 0 0 0· 4 73045 23573 12· 0 0R 0T 0 3498607R 4868605T 0· 1 0R 0T 0 3012246R 3270261T 0· 2 0R 0T 0 2498608R 2839104T 0· 3 0R 0T 0 0R 1983947T 0· 1486579L 0O 1380614Y 2945N 2945F 2734A
workingset_refault_anon 0 workingset_refault_file 18130598
total used free shared buff/cache availableMem: 31978 6705 312 20 24960 24786 Swap: 31977 4 31973
RFC:
Execution Results after 908 seconds
Executed Time (µs) RateSTOCK_LEVEL 2252 27159962888.2 0.08 txn/s
TOTAL 2252 27159962888.2 0.08 txn/s
workingset_refault_anon 22585 workingset_refault_file 22715256
memcg 66 /system.slice/docker-0989446ff78106e32d3f400a0cf371c9a703281bded86d6d6bb1af706ebb25da.scope node 0 22 563007 2274 1198225· 0 0r 1e 0p 0r 697076e 0p 1 0r 0e 0p 0r 0e 325661p 2 0r 0e 0p 0r 0e 888728p 3 0r 0e 0p 0r 0e 3602238p 0 0 0 0 0 0· 23 532222 7525 4948747· 0 0 0 0 0 0 0· 1 0 0 0 0 0 0· 2 0 0 0 0 0 0· 3 0 0 0 0 0 0· 0 0 0 0 0 0· 24 500367 1214667 3292· 0 0 0 0 0 0 0· 1 0 0 0 0 0 0· 2 0 0 0 0 0 0· 3 0 0 0 0 0 0· 0 0 0 0 0 0· 25 469692 40797 466· 0 0R 271T 0 0R 1162165T 0· 1 0R 0T 0 774028R 1205332T 0· 2 0R 0T 0 0R 932484T 0· 3 0R 1T 0 0R 4252158T 0· 25178380L 156515O 23953602Y 59234N 49391F 48664A
total used free shared buff/cache availableMem: 31978 6968 338 5 24671 24555 Swap: 31977 1533 30444
Using same mongodb config (a 3 replica cluster using the same config): { "net": { "bindIpAll": true, "ipv6": false, "maxIncomingConnections": 10000, }, "setParameter": { "disabledSecureAllocatorDomains": "*" }, "replication": { "oplogSizeMB": 10480, "replSetName": "issa-tpcc_0" }, "security": { "keyFile": "/data/db/keyfile" }, "storage": { "dbPath": "/data/db/", "syncPeriodSecs": 60, "directoryPerDB": true, "wiredTiger": { "engineConfig": { "cacheSizeGB": 5 } } }, "systemLog": { "destination": "file", "logAppend": true, "logRotate": "rename", "path": "/data/db/mongod.log", "verbosity": 0 } }
The test environment have 32g memory and 16 core.
Per my analyze, the access pattern for the mongodb test is that page will be re-access long after it's evicted so PID controller won't protect higher tier. That RFC will make use of the long existing shadow to do feedback to PID/Gen so the result will be much better. Still need more adjusting though, will try to do a rebase on top of mm-unstable which includes your patch.
I've no idea why the workingset_refault_* is higher in the better case, this a clearly an IO bound workload, Memory and IO is busy while CPU is not full...
I've uploaded my local reproducer here: https://github.com/ryncsn/emm-test-project/tree/master/mongo-cluster https://github.com/ryncsn/py-tpcc
Thanks for the repos -- I'm trying them right now. Which MongoDB version did you use? setup.sh didn't seem to install it.
Also do you have a QEMU image? It'd be a lot easier for me to duplicate the exact environment by looking into it.