[B! tcp] hiroyukimã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯

hiroyukim id:hiroyukim

tcpã«é–¢ã™ã‚‹hiroyukimã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ (11)

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

${{author_name}}$
{{author_name}}{{created}}
{{ #comment }}{{ comment }}{{ /comment }}
- {{ label }}

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

TIME_WAITã«é–¢ã™ã‚‹è©± | GREE Engineering
ã“ã‚“ã«ã¡ã‚ã€‚ã›ã˜ã¾ã§ã™ã€‚ å…ˆæ—¥ãƒãƒã£ãŸ C302CA ãŒå±Šã„ãŸã‚“ã§ã™ãŒã€ï¼ˆè©±ãŒé•·ããªã‚‹ã®ã§ã™ã”ã„é›‘ã«ã„ã†ã¨ï¼‰Chromebook ã®ã€ã„ã‚ã„ã‚å‰ŠãŽè½ã¨ã—ã¦æœ€é©åŒ–ã™ã‚‹ã¨ã„ã†æ–¹å‘æ€§ã«ã¯ã€ã¨ã¦ã‚‚å¦ã¶ã¨ã“ã‚ãŒã‚ã‚Šã¾ã—ãŸã€‚ æ˜¨å¹´æœ«ã€ MySQLã®ã‚µãƒ¼ãƒé›†ç´„ã«å–ã‚Šçµ„ã‚“ã§ã„ã‚‹ã¨ã„ã†è©±ã‚’ã•ã›ã¦ã„ãŸã ãã¾ã—ãŸãŒã€ã€ŒDBé›†ç´„ã§ãã¦ããŸã®ã§ã€Webã‚¢ãƒ—ãƒªã‚±ãƒ¼ã‚·ãƒ§ãƒ³ã‚µãƒ¼ãƒã‚‚é›†ç´„ã™ã‚‹ã‹ãƒ¼ã€ã¨ã„ã†ã“ã¨ã§ã€æœ€è¿‘ã¯Webã‚µãƒ¼ãƒã®é›†ç´„ã«ã‚‚å–ã‚Šçµ„ã‚“ã§ã„ã¾ã™ã€‚ ãŸã ã€PHPã‚´ãƒªã‚´ãƒªæ›¸ã„ã¦ã‚‹ã®ã§ã¯ãªãã€é›†ç´„ã—ã¦ã„ãä¸Šã§ Linux ã‚„ TCP çš„ãªè¦³ç‚¹ã‹ã‚‰ãƒãƒ¥ãƒ¼ãƒ‹ãƒ³ã‚°ã—ãŸã»ã†ãŒè‰¯ã„ã¨ã“ã‚ã‚’è¦‹ã¦ã¾ã—ã¦ã€ãã®ã¸ã‚“ã«ã¤ã„ã¦ã‚ã‚‹ã¦ã„ã©ã¾ã¨ã¾ã£ãŸã®ã§ã€æ¬¡ã®ã‚¹ãƒ©ã‚¤ãƒ‰ã‚’æ›¸ã‹ã›ã¦ã„ãŸã ãã¾ã—ãŸã€‚
hiroyukim 2017/06/02
TCP

performance
ãƒªãƒ³ã‚¯
TCP Fast Openã®é—‡ã¨ã€Kernelã®ç·©å’Œã‚³ãƒŸãƒƒãƒˆ - ASnoKaze blog
TCP Fast Open TCP Fast Openã¨å‘¼ã°ã‚Œã‚‹æŠ€è¡“ãŒã‚ã‚Šã€RFC 7413ã¨ã—ã¦æ¨™æº–åŒ–ã•ã‚Œã¦ã„ã‚‹ã€‚ ã“ã®TCP Fast Openã‚’ä½¿ã†ã¨ã€ä¸€åº¦ã‚³ãƒã‚¯ã‚·ãƒ§ãƒ³ã‚’è²¼ã£ãŸç›¸æ‰‹ã¨ã¯ã€TCPã®3ã‚¦ã‚§ã‚¤ãƒãƒ³ãƒ‰ã‚·ã‚§ã‚¤ã‚¯ä¸ã«ãƒ‡ãƒ¼ã‚¿ã‚’é€å—ä¿¡ã§ãã‚‹ã‚ˆã†ã«ãªã‚‹ã€‚ã‚¯ãƒ©ã‚¤ã‚¢ãƒ³ãƒˆã‹ã‚‰SYNã¨ã¨ã‚‚ã«ãƒ‡ãƒ¼ã‚¿ã‚’é€ä¿¡ã™ã‚‹ã“ã¨ã§ã€å®Ÿéš›ã«ãƒ‡ãƒ¼ã‚¿ã‚’é€å—ä¿¡é–‹å§‹ã™ã‚‹ã¾ã§ã®å¾…ã¡æ™‚é–“ãŒçŸç¸®ã§ãã‚‹ã€‚ Linuxã§ã¯ã™ã§ã«ã‚¯ãƒ©ã‚¤ã‚¢ãƒ³ãƒˆ/ã‚µãƒ¼ãƒä¸¡æ–¹ã§TCP Fast Openã‚’ä½¿ç”¨ã§ãã‚‹ã€‚ TCP Fast Openã®é—‡ ã—ã‹ã—ã€æ•°å¹´å‰ã‚ˆã‚Šã“ã®TCP Fast Openã«ã¯ä¸€éƒ¨ã®ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã§å¥‡å¦™ãªæŒ¯ã‚‹èˆžã„ã‚’ã™ã‚‹ã“ã¨ãŒçŸ¥ã‚‰ã‚Œã¦ã„ã‚‹ã€‚Appleã®äººãŒå®Ÿéš›ã«ãƒ‡ãƒ—ãƒã‚¤ã—ãŸæ™‚ã«è¦‹ã¤ã‘ãŸã‚‚ã®ã§ã€IETFã‚„NANOGã«ã¦å ±å‘Šã•ã‚Œã¦ãŠã‚Šã€ãã®æ™‚ã®è³‡æ–™ã¯ä¸‹è¨˜ã®ã¨ãŠã‚Šã§ã‚ã‚‹ Deploying TCP Fast Open in the wild
hiroyukim 2017/05/13
network

TCP
ãƒªãƒ³ã‚¯
How TCP backlog works in Linux
When an application puts a socket into LISTEN state using the listen syscall, it needs to specify a backlog for that socket. The backlog is usually described as the limit for the queue of incoming connections. Because of the 3-way handshake used by TCP, an incoming connection goes through an intermediate state SYN RECEIVED before it reaches the ESTABLISHED state and can be returned by the accept s
hiroyukim 2015/06/03
Socket

performance

linux

listen

tcp

queue
ãƒªãƒ³ã‚¯
GitHub - robertswiecki/intrace: Enumeration of IP hops using existing TCP connections
==================================== InTrace is traceroute-like application that enables users to enumerate IP hops using existing TCP connections, both initiated from local network (local system) or from remote hosts. It could be useful for network reconnaissance and firewall bypassing. /* This program is free software; you can redistribute it and/or modify it under the terms of the GNU General P
hiroyukim 2015/03/16
network

tcp

System
ãƒªãƒ³ã‚¯
[ãƒ¡ãƒ¢] TCPä¸Šï¼ˆã‚‚ã—ãã¯HTTPï¼‰ã«ãƒªãƒˆãƒ©ã‚¤å¯èƒ½ãªã‚¢ãƒ—ãƒªã‚±ãƒ¼ã‚·ãƒ§ãƒ³ãƒ—ãƒãƒˆã‚³ãƒ«ã‚’å®Ÿç¾ã™ã‚‹æ–¹æ³•
HTTP/1.1ã®æŒç¶šçš„æŽ¥ç¶šã«ãŠã„ã¦ã¯ã€ã‚µãƒ¼ãƒãŒãƒªã‚¯ã‚¨ã‚¹ãƒˆã‚’å—ã‘å–ã£ãŸã‚ã¨ã«ç•°å¸¸çµ‚äº†ã—ãŸã®ã‹ã€ãƒªã‚¯ã‚¨ã‚¹ãƒˆã‚’å—ã‘å–ã‚‰ãšã«æŽ¥ç¶šã‚’é–‰ã˜ãŸã®ã‹åˆ¤åˆ¥ã™ã‚‹ã“ã¨ãŒã§ããªã„ã€‚ã“ã®ãŸã‚ã€ã¹ãç‰æ€§ã®ä¿è¨¼ãŒãªã„ã‚¢ãƒ—ãƒªã‚±ãƒ¼ã‚·ãƒ§ãƒ³ã«ãŠã„ã¦ã€ãƒªãƒˆãƒ©ã‚¤ã‚’è¡Œã†ã¹ãã‹å¦ã‹è‡ªå‹•çš„ã«åˆ¤æ–ã§ããªããªã‚‹å ´åˆãŒã—ã°ã—ã°ç™ºç”Ÿã™ã‚‹æ³¨ã€‚ ãƒªãƒˆãƒ©ã‚¤å¯èƒ½ã‹å¦ã‹ï¼ˆãƒ”ã‚¢ãŒãƒ¡ãƒƒã‚»ãƒ¼ã‚¸ã®å‡¦ç†ã‚’é–‹å§‹ã—ãŸå¦ã‹ï¼‰ã‚’åˆ¤åˆ¥ã™ã‚‹ã«ã¯ã€ã‚ˆã‚Šç´°ã‹ãªæƒ…å ±äº¤æ›ã‚’è¡Œã†åˆ¥ç¨®ã®ãƒ—ãƒãƒˆã‚³ãƒ«ã‚’æŽ¡ç”¨ã—ã¦ã‚‚è‰¯ã„ãŒã€è¤‡é›‘ãªãƒ—ãƒˆãƒã‚³ãƒ«ã¯ãƒ‘ãƒ•ã‚©ãƒ¼ãƒžãƒ³ã‚¹ã«æ‚ªå½±éŸ¿ã‚’åŠã¼ã™å¯èƒ½æ€§ãŒé«˜ã„ã®ã§é¿ã‘ãŸã„ã¨ã“ã‚ã§ã‚ã‚‹ã€‚ ã¨ã„ã†ã‚ã‘ã§ã€ä»¥ä¸‹æœ¬é¡Œã€‚ pipeliningã‚’è¡Œã‚ãªã„HTTP/1.1ã®ã‚ˆã†ãªå˜ç´”ãªãƒªã‚¯ã‚¨ã‚¹ãƒˆï¼ãƒ¬ã‚¹ãƒãƒ³ã‚¹åž‹ãƒ—ãƒãƒˆã‚³ãƒ«ã‚’ãã®ã¾ã¾ã«ã€ã‚¢ãƒ—ãƒªã‚±ãƒ¼ã‚·ãƒ§ãƒ³ãƒ¬ã‚¤ãƒ¤ã¸ã®ãƒªã‚¯ã‚¨ã‚¹ãƒˆåˆ°é”å¯å¦ã‚’åˆ¤å®šã™ã‚‹æ‰‹è»½ãªæ–¹æ³•ã¨ã—ã¦ã¯ã€SO_LINGERã‚’ç”¨ã„ã‚‹æ–¹æ³•ãŒã‚ã‚‹ã€‚å…·ä½“çš„ã«ã¯ã€ä»¥ä¸‹ã®ã‚ˆã†ãªå½¢å¼ã§ã‚µãƒ¼ãƒã‚’å®Ÿè£…
hiroyukim 2015/01/15
HTTP

kazuho

tcp
ãƒªãƒ³ã‚¯
Data Center Transmission Control Protocol (DCTCP)
Applies To: Windows Server 2012 Data Center Transmission Control Protocol (DCTCP) Data centers host diverse applications, mixing on the same network a variety of workflows that require small, predictable latency, while other applications require large, sustained throughput. In this environment, todayâ€™s state-of-the-art Transmission Control Protocol (TCP) congestion control mechanisms do not provid
hiroyukim 2014/09/28
ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯

ãƒ‘ã‚±ãƒƒãƒˆ

tcp
ãƒªãƒ³ã‚¯
linux ã® TCP_DEFER_ACCEPT (ã‚µãƒ¼ãƒã‚µã‚¤ãƒ‰) ã®æŒ™å‹•ã«ã¤ã„ã¦ - kazuhoã®ãƒ¡ãƒ¢ç½®ãå ´
ä»¥å‰ (2.6.31 ã¾ã§?) ã¯ä»¥ä¸‹ã®æŒ™å‹•*1ã€‚ æœ€åˆã®ãƒšã‚¤ãƒãƒ¼ãƒ‰ã‚’å—ä¿¡ã™ã‚‹ã¾ã§ SYN_RECV ã‚¹ãƒ†ãƒ¼ãƒˆ ã‚¯ãƒ©ã‚¤ã‚¢ãƒ³ãƒˆã® ACK (TCP ãƒãƒ³ãƒ‰ã‚·ã‚§ã‚¤ã‚¯ã®æœ€å¾Œã®ãƒ‘ã‚±ãƒƒãƒˆ) ã‚’å—ä¿¡ã—ã¦ã„ãŸã¨ã—ã¦ã‚‚ã€SYN-ACK ã‚’é€ã‚Šç¶šã‘ã‚‹ 190 ç§’ãŸã£ãŸã‚‰ã€ã‚µãƒ¼ãƒå´ã¯ TCP æŽ¥ç¶šç¢ºç«‹å¤±æ•—ã¨èªè˜ ã‚¯ãƒ©ã‚¤ã‚¢ãƒ³ãƒˆã¯ SYN-ACK ã‚’é€ã£ã¦ã‚‹ã‹ã‚‰æŽ¥ç¶šã§ãã¦ã‚‹ã¤ã‚‚ã‚Š TCP ã®ä»•æ§˜çš„ã«ã©ã†ãªã®ã€ã£ã¦è©±ã¯ã‚ã‹ã‚‹ã€‚ãŸã ã€IP å±¤ã¯ãƒ‘ã‚±ãƒƒãƒˆãƒã‚¹ã®å¯èƒ½æ€§ãŒã‚ã‚‹ã‚ã‘ã§ã€ã‚¤ãƒ³ã‚¿ãƒ¼ãƒãƒƒãƒˆã‚’ä½¿ã£ã¦ã„ã¦ã€ã“ã®æŒ™å‹•ã§å•é¡ŒãŒã‚ã‚‹ã¨ã—ãŸã‚‰ã€ãã‚Œã¯ã‚¢ãƒ—ãƒªã‚±ãƒ¼ã‚·ãƒ§ãƒ³ã®ãƒã‚°ã ã¨æ€ã†ã‘ã©ã€‚ä¸€æ–¹ã§ã€ LAN ä¸Šã§ãƒ‘ã‚±ãƒƒãƒˆãƒã‚¹ãŒ (ã»ã¼) èµ·ã“ã‚‰ãªã„å‰æã§ä½œã£ã¦ãŸã‚‰å›°ã‚‹ã“ã¨ãŒã‚ã‚‹ã®ã‹ãªãƒ¼ã€‚Ubuntu ã® BTS ã§è©±ãŒå‡ºã¦ã‚‹ã®ã¯ã€ãã†ã„ã†ã‚±ãƒ¼ã‚¹ (ç‰¹å®šã®ãƒãƒ¼ãƒ‰ã‚¦ã‚§ã‚¢ãƒãƒ¼ãƒ‰ãƒãƒ©ãƒ³ã‚µã¨ TCP_DEFER_ACCEPT ã‚’ä½¿ã† Apac
hiroyukim 2014/05/12
tcp

accept

linux

server
ãƒªãƒ³ã‚¯
Linuxã§TCP_DEFER_ACCEPTãŒæœ‰åŠ¹ã§ã‚‚acceptå¾Œreadã§ããªã„ç†ç”±
listen()ã®backlogãŒä¸è¶³ã—ãŸéš›ã®TCP_DEFER_ACCEPTã®å‹•ä½œã«ã¤ã„ã¦ - blog.nomadscafe.jpã¨ã„ã†è¨˜äº‹ã®ä¸ã§ã€listen backlog ãŒã‚ãµã‚ŒãŸå¾Œã« accept(2) ã™ã‚‹ã¨ã€ãã®å¾Œã® read(2) ãŒ EAGAIN ã‚’è¿”ã—ãŸã‚Šã€æŽ¥ç¶šãŒä¸å®‰å®šã«ãªã‚‹ã¨ã„ã†äº‹è±¡ãŒèª¬æ˜Žã•ã‚Œã¦ã„ã¾ã—ãŸã€‚æ°—ã«ãªã£ãŸã®ã§èª¿ã¹ã¦ã¿ãŸã“ã¨ã‚’ã¾ã¨ã‚ã¾ã™ã€‚ çµè«–ã‹ã‚‰è¨€ã†ã¨ã“ã‚Œã¯Linuxã®ä»•æ§˜ã§ã™ã€‚manã®tcp(7)ã‚’è¦‹ã‚‹ã¨ã€ TCP_DEFER_ACCEPT (since Linux 2.4) Allow a listener to be awakened only when data arrives on the socket. Takes an integer value (seconds), this can bound the maximum number of
hiroyukim 2014/05/08
accept

tcp
ãƒªãƒ³ã‚¯
SYN cookies - Wikipedia
SYN cookies (ã‚¹ã‚£ãƒ³ãƒ»ã‚¯ãƒƒã‚ãƒ¼ã‚º) ã¨ã¯ã€TCP SYN floodæ”»æ’ƒã‚’é˜²ããŸã‚ã«é–‹ç™ºã•ã‚ŒãŸæ‰‹æ³•ã®ã²ã¨ã¤ã€‚1996å¹´ã€ãƒ€ãƒ‹ã‚¨ãƒ«ãƒ»Jãƒ»ãƒãƒ¼ãƒ³ã‚¹ã‚¿ã‚¤ãƒ³ã‚‰ã«ã‚ˆã‚Šè€ƒæ¡ˆã•ã‚ŒãŸã€‚ SYN flood æ”»æ’ƒã®å•é¡Œç‚¹ã¯ã€TCP æŽ¥ç¶šãŒé–‹å§‹ã™ã‚‹å‰ã‹ã‚‰ã‚µãƒ¼ãƒãŒã‚¯ãƒ©ã‚¤ã‚¢ãƒ³ãƒˆã® SYN ãƒ‘ã‚±ãƒƒãƒˆã«ã‚ˆã£ã¦è¨˜æ†¶é ˜åŸŸã‚’æ¶ˆè²»ã—ã¦ã—ã¾ã†ç‚¹ã«ã‚ã£ãŸã€‚é€šå¸¸ã“ã®è¨˜æ†¶é ˜åŸŸã«ã¯ã€ã‚¯ãƒ©ã‚¤ã‚¢ãƒ³ãƒˆå´ã®IPã‚¢ãƒ‰ãƒ¬ã‚¹ã¨ãƒãƒ¼ãƒˆç•ªå·ã€æŽ¥ç¶šã«ä½¿ã†ã‚·ãƒ¼ã‚±ãƒ³ã‚¹ç•ªå·ã€ãŠã‚ˆã³ã‚¯ãƒ©ã‚¤ã‚¢ãƒ³ãƒˆãŒæŒ‡å®šã—ã¦ããŸ TCPæŽ¥ç¶šã«é–¢ã™ã‚‹æ§˜ã€…ãªè¨å®š (TCPã‚¦ã‚£ãƒ³ãƒ‰ã‚¦ã®å¤§ãã•ãªã©) ãŒæ ¼ç´ã•ã‚Œã‚‹ã€‚ã¾ãŸã€ã‚µãƒ¼ãƒã¯ SYN ãƒ‘ã‚±ãƒƒãƒˆã‚’å—ã‘ã¨ã£ãŸå¾Œã€ã‚¯ãƒ©ã‚¤ã‚¢ãƒ³ãƒˆã«å¯¾ã—ã¦ SYN ACK ãƒ‘ã‚±ãƒƒãƒˆã‚’è¿”ã™ã€‚ã“ã“ã«ã¯ãã® TCP æŽ¥ç¶šã«é–¢é€£ã¥ã‘ã‚‰ã‚ŒãŸTCPã‚·ãƒ¼ã‚±ãƒ³ã‚¹ç•ªå·ãŒå«ã¾ã‚Œã¦ã„ã‚‹ã€‚TCPã‚·ãƒ¼ã‚±ãƒ³ã‚¹ç•ªå·ã¯ã“ã‚Œä»¥é™ã® TCPé€šä¿¡ã®ä¸ã§ã€ã‚¯ãƒ©ã‚¤ã‚¢ãƒ³ãƒˆãŠã‚ˆã³ã‚µãƒ¼ãƒãŒå…±é€šã—ã¦ä½¿
hiroyukim 2013/08/14
tcp

linux

SYN Flood
ãƒªãƒ³ã‚¯
Take advantage of TCP/IP options to optimize data transmission | TechRepublic
Streamlining data transmission is a critical challenge for any enterprise. Thankfully, the TCP/IP protocol provides many ways to optimize data transfer. This article examines several options and shows how they can boost performance. By Alex Kuznetsov, Alex Plant, and Alexander Tormasov Last time, we explained how the TCP_CORK option can decrease the number of packets transf erred over a network. Wh
hiroyukim 2013/08/13
tcp

Linux

performance
ãƒªãƒ³ã‚¯
TCP 3-way handshakeã®ãƒ¬ã‚¤ãƒ†ãƒ³ã‚·è»½æ¸›ã®ãŸã‚ã®TCP_DEFER_ACCEPTã‚½ã‚±ãƒƒãƒˆã‚ªãƒ—ã‚·ãƒ§ãƒ³ - ã‚†ã†ã†ããƒ–ãƒã‚°
2013/10/22 è¿½è¨˜ã—ãŸï¼Ž Starletã®ã‚³ãƒ¼ãƒ‰èªã‚“ã§ã¦listening socketã«TCP_DEFER_ACCEPTã¨ã‹ã„ã†ã‚ªãƒ—ã‚·ãƒ§ãƒ³æ¸¡ã—ã¦ãŸã®ã§ã€ã“ã‚Œä½•ã ã£ã¦æ€ã£ã¦èª¿ã¹ãŸï¼Ž TCPã«ç‰¹ã«è©³ã—ã„ã‚ã‘ã§ã¯ãªã„ã®ã§ç†è§£ã«èª¤ã‚ŠãŒã‚ã‚‹ã‹ã‚‚ã—ã‚Œãªã„ï¼Ž package Starlet::Server; ... # set defer accept if ($^O eq 'linux') { setsockopt($self->{listen_sock}, IPPROTO_TCP, 9, 1) # 9ãŒTCP_DEFER_ACCEPTã‚’è¡¨ã™ and $self->{_using_defer_accept} = 1; } ... TCP_DEFER_ACCEPTã¯Linux 2.4ã‹ã‚‰å°Žå…¥ã•ã‚Œã¦ã„ã‚‹ï¼Ž Linux 2.6.32ã‹ã‚‰æŒ™å‹•ãŒè‹¥å¹²å¤‰ã‚ã£ã¦ã„ã‚‹ã‚‰ã—ã„ï¼Ž (linux ã® TCP_DE
hiroyukim 2013/07/22
Linux

starlet

network

tcp
ãƒªãƒ³ã‚¯
1