VMware上のCentOSでApacheのベンチを行ってみるテスト第二弾
今回は iptables や httpd.conf を出来るだけスリムにしてみたり、実際の運営時の状態にしてみたり、さらに重くしてみたり。
iptablesが限界っぽい
昨日は気が付かなかったが、サーバーログに
Mar 23 03:14:58 localhost kernel: printk: 299 messages suppressed.
Mar 23 03:14:58 localhost kernel: ip_conntrack: table full, dropping packet.
こんなのが沢山残っていた。1万回アクセスはiptabkesでもパケットが捌ききれていない模様。
まあいい。頭の片隅に入れて、同じ環境でテスト。
ベンチ方法
ab -n 10000 -c 300 http://192.168.xx.xx/xxxx.html
ベンチ前後でTOP や ps aux | sort -nk 4 でメモリ使用量やロードアベレージをチェック。
iptablesでアクセス制御
ポート80番をipベース・hashlimitで制御
-A RH-Firewall-1-INPUT -p tcp -m state --state NEW --dport 80-mhashlimit --hashlimit-name t_http --hashlimit 60/m --hashlimit-burst 12--hashlimit-mode srcip --hashlimit-htable-expire 100 -j ACCEPT
とかやってみる
ConcurrencyLevel: 300
Time taken for tests: 12.745237 seconds
Complete requests: 10000
Failedrequests: 9270
(Connect: 0, Length: 9270, Exceptions: 0)
Writeerrors: 0
Total transferred: 420864306 bytes
HTMLtransferred: 418584306 bytes
Requests per second: 784.61 [#/sec] (mean)
Time perrequest: 382.357 [ms] (mean)
Time perrequest: 1.275 [ms] (mean, across all concurrent requests)
Transferrate: 32247.34 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 14222.0 0 9000
Processing: 1 2281267.8 70 12727
Waiting: 0 228 1267.7 69 12726
Total: 56 243 1293.3 70 12744
Percentage of the requests served within a certain time (ms)
50% 70
66% 72
75% 77
80% 78
90% 96
95% 310
98% 866
99% 6318
100% 12744 (longest request)
ほとんどのレスポンスが70ms前後。変わらん。
今回はアタックするIPは1個だけ。本番環境では数千単位。この結果は鵜呑みに出来ない。
(一応、今の所本番環境でもレスポンス低下は見られない)
iptablesでアクセス制御やめる
普通に
-A RH-Firewall-1-INPUT -m state --state NEW -m tcp -p tcp --dport80 -j ACCEPT
って戻してやる
Concurrency Level: 300
Time taken for tests: 13.361442 seconds
Complete requests: 10000
Failedrequests: 9258
(Connect: 0, Length: 9258, Exceptions: 0)
Writeerrors: 0
Total transferred: 420851903 bytes
HTMLtransferred: 418571903 bytes
Requests per second: 748.42 [#/sec] (mean)
Time perrequest: 400.843 [ms] (mean)
Time perrequest: 1.336 [ms] (mean, across all concurrent requests)
Transferrate: 30759.26 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 12184.6 0 3001
Processing: 1 2721438.7 79 13324
Waiting: 0 270 1438.8 78 13323
Total: 29 284 1462.1 80 13360
Percentage of the requests served within a certain time (ms)
50% 80
66% 81
75% 82
80% 83
90% 86
95% 95
98% 1536
99% 13289
100% 13360 (longest request)
何故か数値が落ちた。iptablesも、何度も再起動するとレスポンス落ちるのだろうか?
良好な通信状態の部分(この場合アクセスの90%)のレスポンスが10ms落ちた。
iptablesの設定をドンドン簡略化して行っても、レスポンスが戻らなかったので、再起動直後の状態には戻らないって事だろう
apacheにmod_deflate追加
mod_deflateでHTMLを圧縮転送するようにしてみた。
ConcurrencyLevel: 300
Time taken for tests: 12.753278 seconds
Complete requests: 10000
Failedrequests: 9284
(Connect: 0, Length: 9284, Exceptions: 0)
Writeerrors: 0
Total transferred: 421280555 bytes
HTMLtransferred: 418660031 bytes
Requests per second: 784.11 [#/sec] (mean)
Time perrequest: 382.598 [ms] (mean)
Time perrequest: 1.275 [ms] (mean, across all concurrent requests)
Transferrate: 32258.84 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 11245.3 0 9000
Processing: 6 2691411.4 70 12734
Waiting: 5 268 1411.3 70 12733
Total: 6 280 1444.9 71 12751
Percentage of the requests served within a certain time (ms)
50% 71
66% 78
75% 80
80% 82
90% 84
95% 87
98% 2467
99% 12699
100% 12751 (longest request)
もう一回行ったら以下な感じ
50% 80
66% 82
75% 83
80% 84
90% 86
95% 91
98% 1347
99% 3699
100% 6371 (longest request)
レスポンス70~80の間は誤差か?
DEFLATE導入だけじゃ変わらない。
▼ちなみにApacheに追加しかのは以下。
SetOutputFilterDEFLATE
BrowserMatch^Mozilla/4 gzip-only-text/html
BrowserMatch^Mozilla/4\.0[678] no-gzip
BrowserMatch \bMSI[E]!no-gzip !gzip-only-text/html
SetEnvIfNoCaseRequest_URI \.(?:gif|jpe?g|png|ico|zip|avi|mp4)$ no-gzip dont-vary
Header append VaryUser-Agent env=!dont-vary
で、これ apachectl -k graceful や apachectl restart では圧縮設定が反映されず。
apachectl stop apachectl start が必要だった。
移行、Apache再起動は apachectl stop apachectl start を使う事に。
order allow,deny設定その1
SetEnvIfを複数行使ってUAやリファラ拒否
deny from xxx.xxx.xxx.xxx
deny from env=xxxxx
を複数行、大量に用意(約200行)
ConcurrencyLevel: 300
Time taken for tests: 12.757112 seconds
Complete requests: 10000
Failedrequests: 9300
(Connect: 0, Length: 9300, Exceptions: 0)
Writeerrors: 0
Total transferred: 421201539 bytes
HTMLtransferred: 418581539 bytes
Requests per second: 783.88 [#/sec] (mean)
Time perrequest: 382.713 [ms] (mean)
Time perrequest: 1.276 [ms] (mean, across all concurrent requests)
Transferrate: 32243.11 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 9 184.7 0 8999
Processing: 3 2621385.8 64 12743
Waiting: 3 262 1385.7 63 12741
Total: 3 272 1405.8 64 12751
Percentage of the requests served within a certain time (ms)
50% 64
66% 72
75% 77
80% 79
90% 83
95% 128
98% 2248
99% 12698
100% 12751 (longest request)
[root@localhost ~]#
なんか今までで最高のレスポンスがw
実はここで、一旦NMwareを完全停止、再起動していた。
allow denyもあんまり関係ないのかな?
apache 3599 0.3 1.7 312412 9148? S 03:35 0:00 /usr/sbin/httpd -k start
apache 3600 0.5 1.7 312268 8876? S 03:35 0:01 /usr/sbin/httpd -k start
apache 3603 0.4 1.7 312268 8876? S 03:35 0:00 /usr/sbin/httpd -k start
apache 3604 0.3 1.7 312268 8876? S 03:35 0:00 /usr/sbin/httpd -k start
apache 3607 0.2 1.7 312268 8876? S 03:36 0:00 /usr/sbin/httpd -k start
apache 3608 0.1 1.7 312268 8876? S 03:36 0:00 /usr/sbin/httpd -k start
apache 3609 0.1 1.7 312268 8876? S 03:36 0:00 /usr/sbin/httpd -k start
apache 3610 0.0 1.7 312268 8876? S 03:36 0:00 /usr/sbin/httpd -k start
apache 3611 0.0 1.7 312268 8876? S 03:36 0:00 /usr/sbin/httpd -k start
apache 3612 0.0 1.7 312268 8876? S 03:36 0:00 /usr/sbin/httpd -k start
apache 3613 0.0 1.7 312268 8876? S 03:36 0:00 /usr/sbin/httpd -k start
apache 3614 0.0 1.7 312268 8876? S 03:36 0:00 /usr/sbin/httpd -k start
apache 3616 0.0 1.7 312268 8876? S 03:36 0:00 /usr/sbin/httpd -k start
apache 3617 0.0 1.7 312268 8876? S 03:36 0:00 /usr/sbin/httpd -k start
apache 3618 0.0 1.7 312268 8876? S 03:36 0:00 /usr/sbin/httpd -k start
apache 3619 0.0 1.7 312268 8876? S 03:36 0:00 /usr/sbin/httpd -k start
apache 3620 0.0 1.7 312268 8876? S 03:36 0:00 /usr/sbin/httpd -k start
apache 3621 0.0 1.7 312268 8876? S 03:36 0:00 /usr/sbin/httpd -k start
apache 3622 0.0 1.7 312268 8876? S 03:36 0:00 /usr/sbin/httpd -k start
apache 3623 0.0 1.7 312268 8876? S 03:36 0:00 /usr/sbin/httpd -k start
root 3597 0.0 2.4 31200012624? Ss 03:35 0:00 /usr/sbin/httpd -k start
メモリ使用量も変化なし。
OOOOorder allow,deny設定その2
deny from xxx.xxx.xxx.xxxをさらに大量(合計300個)用意
Concurrency Level: 300
Time taken for tests: 17.503074 seconds
Complete requests: 10000
Failedrequests: 9303
(Connect: 0, Length: 9303, Exceptions: 0)
Writeerrors: 0
Total transferred: 421198763 bytes
HTMLtransferred: 418578763 bytes
Requests per second: 571.33 [#/sec] (mean)
Time perrequest: 525.092 [ms] (mean)
Time perrequest: 1.750 [ms] (mean, across all concurrent requests)
Transferrate: 23500.21 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 7 184.7 0 8999
Processing: 4 3572010.0 70 17490
Waiting: 4 357 2009.8 70 17489
Total: 4 365 2034.5 70 17497
Percentage of the requests served within a certain time (ms)
50% 70
66% 75
75% 79
80% 81
90% 84
95% 91
98% 1872
99% 17331
100% 17497 (longest request)
[root@localhost ~]#
変わらないとみて良いだろう。
禁止IP、禁止UAをまとめてみる
SetEnvIfの内容をこの方法で正規表現で括って数行にまとめてみる。IP400個、UA60個位放り込む。
Concurrency Level: 300
Time taken for tests: 12.741683 seconds
Complete requests: 10000
Failed requests: 9309
(Connect: 0, Length: 9309, Exceptions: 0)
Write errors: 0
Total transferred: 421393609 bytes
HTML transferred: 418772299 bytes
Requests per second: 784.83 [#/sec] (mean)
Time per request: 382.250 [ms] (mean)
Time per request: 1.274 [ms] (mean, across all concurrent requests)
Transfer rate: 32296.91 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 11 198.7 0 9001
Processing: 1 234 1328.4 65 12725
Waiting: 0 233 1328.3 65 12723
Total: 4 245 1344.5 66 12741
Percentage of the requests served within a certain time (ms)
50% 66
66% 69
75% 73
80% 78
90% 88
95% 132
98% 813
99% 12682
100% 12741 (longest request
)あれ?サーバー丸ごと再起動していないのに、60ms台になった。
Apacheを数回再起動すると、その後は初回の値に戻ることは今まで無かったのに。。。
最初のテスト時、200行前後はdeny from xxx.xxx.xxx.xxx を入れていたし、、、
deny from xxx.xxx.xxx.xxx 20行よりも環境変数に全部ぶち込んでdeny from env=deny_ua 1回の方が高速なのか?
もう一回 deny from xxx.xxx.xxx.xxx60行くらい追加
Concurrency Level: 300
Time taken for tests: 12.736661 seconds
Complete requests: 10000
Failed requests: 9327
(Connect: 0, Length: 9327, Exceptions: 0)
Write errors: 0
Total transferred: 422206988 bytes
HTML transferred: 419580700 bytes
Requests per second: 785.14 [#/sec] (mean)
Time per request: 382.100 [ms] (mean)
Time per request: 1.274 [ms] (mean, across all concurrent requests)
Transfer rate: 32371.98 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 21 270.8 0 9001
Processing: 1 197 1128.5 66 12723
Waiting: 0 196 1128.4 66 12721
Total: 4 218 1163.6 67 12735
Percentage of the requests served within a certain time (ms)
50% 67
66% 70
75% 77
80% 79
90% 87
95% 139
98% 898
99% 4174
100% 12735 (longest request)
もう一回 禁止IP、禁止UAをまとめてみる
Concurrency Level: 300
Time taken for tests: 20.581883 seconds
Complete requests: 10000
Failed requests: 9301
(Connect: 0, Length: 9301, Exceptions: 0)
Write errors: 0
Total transferred: 421199937 bytes
HTML transferred: 418579937 bytes
Requests per second: 485.86 [#/sec] (mean)
Time per request: 617.456 [ms] (mean)
Time per request: 2.058 [ms] (mean, across all concurrent requests)
Transfer rate: 19984.96 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 4 119.9 0 3002
Processing: 1 243 1252.0 65 20568
Waiting: 1 243 1251.8 64 20566
Total: 4 248 1258.2 65 20581
Percentage of the requests served within a certain time (ms)
50% 65
66% 69
75% 70
80% 71
90% 81
95% 89
98% 1006
99% 8864
100% 20581 (longest request)
[root@localhost ~]#
やっぱりわずかだけど早くなっている?
すげー僅かだけど事前に環境変数にぶち込んで、denyやallowの回数は減らした方が良いかもしんない。
正規表現的をサーバーに優しくない方法で表記
SetEnvIf Remote_Addr "^(4\.78\.1\.36|38\.105\.71\.13|38\.105\.83\.22)$" deny_ua
上記の赤字部分、正規表現の先頭や末尾を表わす部分や、「.」のエスケープを外してワイルドカード扱いにしたり、CPUに優しくない表現にしてみる。リスト500個の全部外すのは大変だったので、全体の7割ほど外して検証。
今までの経験上、かなりレスポンスが悪くなると予想するが、、、
Concurrency Level: 300
Time taken for tests: 11.836919 seconds
Complete requests: 10000
Failed requests: 9335
(Connect: 0, Length: 9335, Exceptions: 0)
Write errors: 0
Total transferred: 421695398 bytes
HTML transferred: 419072254 bytes
Requests per second: 844.81 [#/sec] (mean)
Time per request: 355.108 [ms] (mean)
Time per request: 1.184 [ms] (mean, across all concurrent requests)
Transfer rate: 34790.39 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 8 187.2 0 9000
Processing: 1 264 1355.9 76 11826
Waiting: 0 264 1355.7 75 11825
Total: 3 272 1371.7 76 11833
Percentage of the requests served within a certain time (ms)
50% 76
66% 80
75% 81
80% 82
90% 85
95% 126
98% 1078
99% 11779
100% 11833 (longest request)
僅かながら違いが出た、と見て良いのか? 思ったよりレスポンス落ちなかった。
でもやっぱ正規表現は出来るだけCPUに負荷のかからない方法で書いた方が良いな。
CustomLogを取るようにしてみた
Concurrency Level: 300
Time taken for tests: 12.748234 seconds
Complete requests: 10000
Failed requests: 9312
(Connect: 0, Length: 9312, Exceptions: 0)
Write errors: 0
Total transferred: 421343374 bytes
HTML transferred: 418722326 bytes
Requests per second: 784.42 [#/sec] (mean)
Time per request: 382.447 [ms] (mean)
Time per request: 1.275 [ms] (mean, across all concurrent requests)
Transfer rate: 32276.47 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 10 205.4 0 9001
Processing: 2 259 1333.1 71 12728
Waiting: 1 259 1333.1 70 12728
Total: 4 270 1355.2 71 12745
Percentage of the requests served within a certain time (ms)
50% 71
66% 81
75% 82
80% 83
90% 86
95% 94
98% 1793
99% 12682
100% 12745 (longest request)
1万件、約1.2MBのアクセスログが出来上がったが、レスポンスに変化なし、とみて良いだろう
数回ベンチを繰り返し、CustomLogを肥大化させてみる
ログが3~4MB超えた辺りから、レスポンスが1割落ちた?
Percentage of the requests served within a certain time (ms)
50% 84
66% 86
75% 87
80% 87
90% 89
95% 93
98% 3082
99% 18754
100% 18902 (longest request)
ログ肥大化した状態になると、サーバー丸ごと再起動しても、レスポンス改善は見られない。
ここでサーバーログを止めてみる
50% 69
66% 78
75% 85
80% 87
90% 91
95% 121
98% 818
99% 12642
100% 12698 (longest request)
60ms後半まで復活。
何度かAcpahe再起動、ApacheBench実行で子プロセスが何個も立ち上がった状態。
HDDやI/O部分の性能によるだろうけど、サーバーログを残すことで10~20msは余分に時間がかかると見て良いだろう。
サーバログにenv=追加 1
CustomLog "|/usr/sbin/rotatelogs /var/www/log/xxx 86400" combined env=!nolog な感じに。
nologは画像やdummy connectionのみ。
ベンチのログは全て記録される
Concurrency Level: 300
Time taken for tests: 12.772569 seconds
Complete requests: 10000
Failed requests: 9290
(Connect: 0, Length: 9290, Exceptions: 0)
Write errors: 0
Total transferred: 421275496 bytes
HTML transferred: 418654972 bytes
Requests per second: 782.93 [#/sec] (mean)
Time per request: 383.177 [ms] (mean)
Time per request: 1.277 [ms] (mean, across all concurrent requests)
Transfer rate: 32209.73 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 6 172.2 0 9000
Processing: 2 289 1470.1 82 12758
Waiting: 1 289 1470.0 82 12757
Total: 6 295 1484.2 82 12770
Percentage of the requests served within a certain time (ms)
50% 82
66% 84
75% 86
80% 87
90% 90
95% 129
98% 2083
99% 12706
100% 12770 (longest request)
まあこんなもんだな。
サーバログにenv=追加 2
SetEnvIf deny_ua "1" nolog
CustomLog "|/usr/sbin/rotatelogs /var/www/log/xxx 86400" combined env=!nolog
としてDENYしたホストからのログも残さないように。
ベンチのログは全て記録される
Concurrency Level: 300
Time taken for tests: 12.775212 seconds
Complete requests: 10000
Failed requests: 9278
(Connect: 0, Length: 9278, Exceptions: 0)
Write errors: 0
Total transferred: 421237769 bytes
HTML transferred: 418617507 bytes
Requests per second: 782.77 [#/sec] (mean)
Time per request: 383.256 [ms] (mean)
Time per request: 1.278 [ms] (mean, across all concurrent requests)
Transfer rate: 32200.25 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 9 194.2 0 9001
Processing: 6 290 1493.5 78 12762
Waiting: 5 290 1493.5 77 12762
Total: 6 299 1513.2 78 12772
Percentage of the requests served within a certain time (ms)
50% 78
66% 83
75% 85
80% 86
90% 88
95% 128
98% 2216
99% 12703
100% 12772 (longest request)
SetEnvIfが一個増えた位じゃ何も変わらん。
サーバログにenv=追加 3
SetEnvIf Remote_Addr "^(192\.168\.)" my_ua
SetEnvIf my_ua "1" nolog
CustomLog "|/usr/sbin/rotatelogs /var/www/log/xxx 86400" combined env=!nolog
として、自分のアクセスもログに残らないようにしてみる。
Concurrency Level: 300
Time taken for tests: 20.834012 seconds
Complete requests: 10000
Failed requests: 9310
(Connect: 0, Length: 9310, Exceptions: 0)
Write errors: 0
Total transferred: 421230122 bytes
HTML transferred: 418609860 bytes
Requests per second: 479.98 [#/sec] (mean)
Time per request: 625.020 [ms] (mean)
Time per request: 2.083 [ms] (mean, across all concurrent requests)
Transfer rate: 19744.49 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 7 169.5 0 9000
Processing: 4 404 2386.8 72 20821
Waiting: 4 404 2386.4 72 20819
Total: 4 412 2403.2 72 20828
Percentage of the requests served within a certain time (ms)
50% 72
66% 84
75% 86
80% 87
90% 92
95% 129
98% 1569
99% 20586
100% 20828 (longest request)
若干、というかかなり復活。
やっぱ10~20ms変わる。ログは極力少なくした方が良いな。
おまけ
サーバー起動直後と、ApacheBenchを二回実行しただけの状態でpsコマンドを取ってみる
サーバー起動直後:ab -n 10000 -c 300 前の ps aux | sort -nk 4
apache 3851 0.0 1.4 312120 7316? S 03:44 0:00 /usr/sbin/httpd -k start
apache 3852 0.0 1.4 312120 7312? S 03:44 0:00 /usr/sbin/httpd -k start
apache 3853 0.0 1.4 312120 7312? S 03:44 0:00 /usr/sbin/httpd -k start
apache 3854 0.0 1.4 312120 7312? S 03:44 0:00 /usr/sbin/httpd -k start
apache 3855 0.0 1.4 312120 7312? S 03:44 0:00 /usr/sbin/httpd -k start
apache 3856 0.0 1.4 312120 7312? S 03:44 0:00 /usr/sbin/httpd -k start
apache 3857 0.0 1.4 312120 7312? S 03:44 0:00 /usr/sbin/httpd -k start
apache 3858 0.0 1.4 312120 7312? S 03:44 0:00 /usr/sbin/httpd -k start
root 3848 1.6 2.4 31198412660? Ss 03:44 0:00 /usr/sbin/httpd -k start
ab -n 10000 -c 300 後の ps aux | sort -nk 4
apache 3852 0.3 1.7 312384 8924? S 03:44 0:00 /usr/sbin/httpd -k start
apache 3853 0.3 1.7 312384 8924? S 03:44 0:00 /usr/sbin/httpd -k start
apache 3854 0.3 1.7 312384 8924? S 03:44 0:00 /usr/sbin/httpd -k start
apache 3855 0.3 1.7 312384 8924? S 03:44 0:00 /usr/sbin/httpd -k start
apache 3856 0.2 1.7 312384 8936? S 03:44 0:00 /usr/sbin/httpd -k start
apache 3858 0.3 1.7 312384 8924? S 03:44 0:00 /usr/sbin/httpd -k start
apache 3864 0.1 1.7 312384 8924? S 03:45 0:00 /usr/sbin/httpd -k start
apache 3865 0.2 1.7 312384 8924? S 03:45 0:00 /usr/sbin/httpd -k start
apache 3866 0.1 1.7 312384 8924? S 03:45 0:00 /usr/sbin/httpd -k start
apache 3867 0.0 1.7 312384 8924? S 03:45 0:00 /usr/sbin/httpd -k start
apache 3869 0.0 1.7 312384 8924? S 03:45 0:00 /usr/sbin/httpd -k start
apache 3870 0.2 1.7 312384 8924? S 03:45 0:00 /usr/sbin/httpd -k start
apache 3871 0.2 1.7 312384 8924? S 03:45 0:00 /usr/sbin/httpd -k start
apache 3873 0.0 1.7 312384 8924? S 03:45 0:00 /usr/sbin/httpd -k start
apache 3874 0.0 1.7 312384 8924? S 03:45 0:00 /usr/sbin/httpd -k start
apache 3875 0.0 1.7 312384 8924? S 03:45 0:00 /usr/sbin/httpd -k start
apache 3876 0.0 1.7 312384 8924? S 03:45 0:00 /usr/sbin/httpd -k start
apache 3877 0.0 1.7 312384 8924? S 03:45 0:00 /usr/sbin/httpd -k start
apache 3878 0.0 1.7 312384 8924? S 03:45 0:00 /usr/sbin/httpd -k start
apache 3879 0.0 1.7 312384 8924? S 03:45 0:00 /usr/sbin/httpd -k start
root 3848 0.0 2.4 31198412668? Ss 03:44 0:00 /usr/sbin/httpd -k start
ApacheBenchのアクセスでもそれなりにメモリ増えるんだねー。
このベンチテスト中、最初からここまでTOPで見るメモリ空き容量は170MB~150MBで全然変わらん。
定期的にapacheを再起動させる事も、サーバー自体を再起動させることも、それなりに意味がありそうだと感じた。
と言っても、apacheの再起動は1~2週間に1回、サーバーの再起動は1~3ヶ月に1回やれば十分な感じだけど。