#dmidecode -t processor # dmidecode 3.0 Getting SMBIOS data from sysfs. SMBIOS 3.2.0 present. # SMBIOS implementations newer than version 3.0 are not # fully supported by this version of dmidecode.
Handle 0x0004, DMI type 4, 48 bytes Processor Information Socket Designation: BGA3576 Type: Central Processor Family: <OUT OF SPEC> Manufacturer: PHYTIUM ID: 00 00 00 00 70 1F 66 22 Version: FT2500 Voltage: 0.8 V External Clock: 50 MHz Max Speed: 2100 MHz Current Speed: 2100 MHz Status: Populated, Enabled Upgrade: Other L1 Cache Handle: 0x0005 L2 Cache Handle: 0x0007 L3 Cache Handle: 0x0008 Serial Number: 1234567 Asset Tag: No Asset Tag Part Number: NULL Core Count: 64 Core Enabled: 64 Thread Count: 64 Characteristics: 64-bit capable Multi-Core Hardware Thread Execute Protection Enhanced Virtualization Power/Performance Control
#lscpu Architecture: aarch64 Byte Order: Little Endian CPU(s): 128 On-line CPU(s) list: 0-127 Thread(s) per core: 1 Core(s) per socket: 64 Socket(s): 2 NUMA node(s): 16 Model: 3 BogoMIPS: 100.00 L1d cache: 32K L1i cache: 32K L2 cache: 2048K L3 cache: 65536K NUMA node0 CPU(s): 0-7 NUMA node1 CPU(s): 8-15 NUMA node2 CPU(s): 16-23 NUMA node3 CPU(s): 24-31 NUMA node4 CPU(s): 32-39 NUMA node5 CPU(s): 40-47 NUMA node6 CPU(s): 48-55 NUMA node7 CPU(s): 56-63 NUMA node8 CPU(s): 64-71 NUMA node9 CPU(s): 72-79 NUMA node10 CPU(s): 80-87 NUMA node11 CPU(s): 88-95 NUMA node12 CPU(s): 96-103 NUMA node13 CPU(s): 104-111 NUMA node14 CPU(s): 112-119 NUMA node15 CPU(s): 120-127 Flags: fp asimd evtstrm aes pmull sha1 sha2 crc32 cpuid
#perf stat -e branch-misses,bus-cycles,cache-misses,cache-references,cpu-cycles,instructions,L1-dcache-load-misses,L1-dcache-loads,L1-dcache-store-misses,L1-dcache-stores,L1-icache-load-misses,L1-icache-loads,branch-load-misses,branch-loads,dTLB-load-misses,iTLB-load-misses -a -p 79694 ^C Performance counter stats for process id '79694':
#第二个MySQL IPC只有第三个的30%多点，这就是为什么CPU高这么多，但是QPS差不多 perf stat -e branch-misses,bus-cycles,cache-misses,cache-references,cpu-cycles,instructions,L1-dcache-load-misses,L1-dcache-loads,L1-dcache-store-misses,L1-dcache-stores,L1-icache-load-misses,L1-icache-loads,branch-load-misses,branch-loads,dTLB-load-misses,iTLB-load-misses -a -p 61238 ^C Performance counter stats for process id '61238':
1.163750756 seconds time elapsed #第三个MySQL perf stat -e branch-misses,bus-cycles,cache-misses,cache-references,cpu-cycles,instructions,L1-dcache-load-misses,L1-dcache-loads,L1-dcache-store-misses,L1-dcache-stores,L1-icache-load-misses,L1-icache-loads,branch-load-misses,branch-loads,dTLB-load-misses,iTLB-load-misses -a -p 53400 ^C Performance counter stats for process id '53400':
#第二个 MySQL IPC只有第三个的30%多点，这就是为什么CPU高这么多，但是QPS差不多 perf stat -e branch-misses,bus-cycles,cache-misses,cache-references,cpu-cycles,instructions,L1-dcache-load-misses,L1-dcache-loads,L1-dcache-store-misses,L1-dcache-stores,L1-icache-load-misses,L1-icache-loads,branch-load-misses,branch-loads,dTLB-load-misses,iTLB-load-misses -a -p 61238 ^C Performance counter stats for process id '61238':
#第三个 MySQL perf stat -e branch-misses,bus-cycles,cache-misses,cache-references,cpu-cycles,instructions,L1-dcache-load-misses,L1-dcache-loads,L1-dcache-store-misses,L1-dcache-stores,L1-icache-load-misses,L1-icache-loads,branch-load-misses,branch-loads,dTLB-load-misses,iTLB-load-misses -a -p 53400 ^C Performance counter stats for process id '53400':
Modern multiprocessor systems mix these basic architectures as seen in the following diagram:
In this complex hierarchical scheme, processors are grouped by their physical location on one or the other multi-core CPU package or “node.” Processors within a node share access to memory modules as per the UMA shared memory architecture. At the same time, they may also access memory from the remote node using a shared interconnect, but with slower performance as per the NUMA shared memory architecture.
Zone_reclaim_mode allows someone to set more or less aggressive approaches to reclaim memory when a zone runs out of memory. If it is set to zero then no zone reclaim occurs. Allocations will be satisfied from other zones / nodes in the system.
zone_reclaim_mode的四个参数值的意义分别是：
0 = Allocate from all nodes before reclaiming memory 1 = Reclaim memory from local node vs allocating from next node 2 = Zone reclaim writes dirty pages out 4 = Zone reclaim swaps pages
Automatic NUMA balancing improves the performance of applications running on NUMA hardware systems. It is enabled by default on Red Hat Enterprise Linux 7 systems.
An application will generally perform best when the threads of its processes are accessing memory on the same NUMA node as the threads are scheduled. Automatic NUMA balancing moves tasks (which can be threads or processes) closer to the memory they are accessing. It also moves application data to memory closer to the tasks that reference it. This is all done automatically by the kernel when automatic NUMA balancing is active.
#perf stat -e sched:sched_stick_numa,sched:sched_move_numa,sched:sched_swap_numa,migrate:mm_migrate_pages,minor-faults -p 7191 Performance counter stats for process id '7191':
0 sched:sched_stick_numa (100.00%) 1 sched:sched_move_numa (100.00%) 0 sched:sched_swap_numa 0 migrate:mm_migrate_pages 286 minor-faults # perf stat -e sched:sched_stick_numa,sched:sched_move_numa,sched:sched_swap_numa,migrate:mm_migrate_pages,minor-faults -p PID ... 1 sched:sched_stick_numa 3 sched:sched_move_numa 41 sched:sched_swap_numa 5,239 migrate:mm_migrate_pages 50,161 minor-faults #perf stat -e sched:sched_stick_numa,sched:sched_move_numa,sched:sched_swap_numa,migrate:mm_migrate_pages,minor-faults -p 676322 Performance counter stats for process id '676322':
答： 能，这个时候在client和server两边看到的连接状态都是 ESTABLISHED，只是Server上的全连接队列占用加1。连接握手并切换到ESTABLISHED状态都是由OS来负责的，应用不参与，ESTABLISHED后应用才能accept，进而收发数据。也就是能放入到全连接队列里面的连接肯定都是 ESTABLISHED 状态的了
接着上面的问题，如果新连接继续连接进而全连接队列满了呢？
答：那就连不上了，server端的OS因为全连接队列满了直接扔掉第一个syn握手包，这个时候连接在client端是SYN_SENT，Server端没有这个连接，这是因为syn到server端就直接被OS drop 了。
由于在 Zen 1 的基础上进行了大量的修改，海光 CPU 可以不用简单地称之为换壳 AMD 处理器了。但其性能相比同代原版 CPU 略差：整数性能基本相同，浮点性能显著降低——普通指令吞吐量只有基准水平的一半。海光 CPU 的随机数生成机制也被修改，加密引擎已被替换，不再对常见的 AES 指令进行加速，但覆盖了其他面向国内安全性的指令如 SM2、SM3 和 SM4。
从以上截图，可以看到关键的 insn per cycle 能到0.51和0.66（这个数值越大性能越好）
如果同时压物理机上的所有服务节点
从以上截图，可以看到关键的 insn per cycle 能降到了0.27和0.31（这个数值越大性能越好），基本相当于单压的5折
通过 perf list 找出所有Hardware event，然后对他们进行perf:
1
sudo perf stat -e branch-instructions,branch-misses,cache-references,cpu-cycles,instructions,stalled-cycles-backend,stalled-cycles-frontend,L1-dcache-load-misses,L1-dcache-loads,L1-dcache-prefetches,L1-icache-load-misses,L1-icache-loads,branch-load-misses,branch-loads,dTLB-load-misses,dTLB-loads,iTLB-load-misses,iTLB-loads -a -- `pidof java`
If you are using the option -XX:+UseSHM or -XX:+UseHugeTLBFS, then specify the number of large pages. In the following example, 3 GB of a 4 GB system are reserved for large pages (assuming a large page size of 2048kB, then 3 GB = 3 * 1024 MB = 3072 MB = 3072 * 1024 kB = 3145728 kB and 3145728 kB / 2048 kB = 1536):
#taskset -c 1,53 /usr/bin/sysbench --num-threads=2 --test=cpu --cpu-max-prime=50000 run sysbench 0.5: multi-threaded system evaluation benchmark
Running the test with following options: Number of threads: 2 Random number generator seed is 0 and will be ignored
Primer numbers limit: 50000
Threads started!
General statistics: total time: 48.5571s total number of events: 10000 total time taken by event execution: 97.0944s response time: min: 8.29ms avg: 9.71ms max: 20.88ms approx. 95 percentile: 9.71ms
Threads fairness: events (avg/stddev): 5000.0000/2.00 execution time (avg/stddev): 48.5472/0.01
#taskset -c 1 /usr/bin/sysbench --num-threads=1 --test=cpu --cpu-max-prime=50000 run sysbench 0.5: multi-threaded system evaluation benchmark
Running the test with following options: Number of threads: 1 Random number generator seed is 0 and will be ignored
Primer numbers limit: 50000
Threads started!
General statistics: total time: 83.2642s total number of events: 10000 total time taken by event execution: 83.2625s response time: min: 8.27ms avg: 8.33ms max: 10.03ms approx. 95 percentile: 8.36ms
Threads fairness: events (avg/stddev): 10000.0000/0.00 execution time (avg/stddev): 83.2625/0.00
#lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 104 On-line CPU(s) list: 0-103 Thread(s) per core: 2 Core(s) per socket: 26 Socket(s): 2 NUMA node(s): 2 Vendor ID: GenuineIntel CPU family: 6 Model: 85 Model name: Intel(R) Xeon(R) Platinum 8269CY CPU @ 2.50GHz Stepping: 7 CPU MHz: 3200.097 CPU max MHz: 3800.0000 CPU min MHz: 1200.0000 BogoMIPS: 4998.89 Virtualization: VT-x L1d cache: 32K L1i cache: 32K L2 cache: 1024K L3 cache: 36608K NUMA node0 CPU(s): 0-25,52-77 NUMA node1 CPU(s): 26-51,78-103 Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf eagerfpu pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch ida arat epb invpcid_single pln pts dtherm spec_ctrl ibpb_support tpr_shadow vnmi flexpriority ept vpid fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm cqm mpx rdt avx512f avx512dq rdseed adx smap clflushopt avx512cdavx512bw avx512vl xsaveopt xsavec xgetbv1 cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local cat_l3 mba
Basically, they are all the same, in the way they all permit the logging of data from different types of systems in a central repository.
But they are three different project, each project trying to improve the previous one with more reliability and functionalities.
The Syslog project was the very first project. It started in 1980. It is the root project to Syslog protocol. At this time Syslog is a very simple protocol. At the beginning it only supports UDP for transport, so that it does not guarantee the delivery of the messages.
Next came syslog-ng in 1998. It extends basic syslog protocol with new features like:
content-based filtering
Logging directly into a database
TCP for transport
TLS encryption
Next came Rsyslog in 2004. It extends syslog protocol with new features like:
# systemd-analyze critical-chain systemd-journald.service The time after the unit is active or started is printed after the "@" character. The time the unit takes to start is printed after the "+" character.
# cat /etc/systemd/journald.conf # This file is part of systemd. # # systemd is free software; you can redistribute it and/or modify it # under the terms of the GNU Lesser General Public License as published by # the Free Software Foundation; either version 2.1 of the License, or # (at your option) any later version. # # Entries in this file show the compile time defaults. # You can change settings by editing this file. # Defaults can be restored by simply deleting this file. # # See journald.conf(5) for details.
-n or –lines= Show the most recent **n** number of log lines
-f or –follow Like a tail operation for viewing live updates
-S, –since=, -U, –until= Search based on a date. “2019-07-04 13:19:17”, “00:00:00”, “yesterday”, “today”, “tomorrow”, “now” are valid formats. For complete time and date specification, see systemd.time(7)
Generally services keep the log files opened while they are running. This mean that they do not care if the log files are renamed/moved or deleted they will continue to write to the open file handled.
When logrotate move the files, the services keep writing to the same file.
Example: syslogd will write to /var/log/cron.log. Then logrotate will rename the file to /var/log/cron.log.1, so syslogd will keep writing to the open file /var/log/cron.log.1.
Sending the HUP signal to syslogd will force him to close existing file handle and open new file handle to the original path /var/log/cron.log which will create a new file.
The use of the HUP signal instead of another one is at the discretion of the program. Some services like php-fpm will listen to the USR1 signal to reopen it’s file handle without terminating itself.
