Comment mesurer simultanément les temps de réponse et de requête à l'aide de cURL?
J'ai un service Web qui reçoit des données au format JSON, les traite, puis renvoie le résultat au demandeur.
Je souhaite mesurer la demande, la réponse et le temps total à l'aide de cURL.
Mon exemple de requête ressemble à:
curl -X POST -d @file server:port
et je mesure actuellement cela en utilisant la commande time sous Linux:
time curl -X POST -d @file server:port
La commande time mesure uniquement total temps, ce qui n’est pas tout à fait ce que je cherche.
Est-il possible de mesurer les temps de réponse et de requête à l'aide de cURL?
Extrait de ce brillant article de blog ... https://blog.josephscott.org/2011/10/14/timing-details-with-curl/
cURL prend en charge la sortie formatée pour les détails de la demande (voir la page de manuel cURL pour plus de détails , sous -w, –write-out <format>). Pour nos besoins, nous nous concentrerons uniquement sur les détails de synchronisation fournis.
Créez un nouveau fichier, curl-format.txt et collez-le dans:
time_namelookup: %{time_namelookup}\n time_connect: %{time_connect}\n time_appconnect: %{time_appconnect}\n time_pretransfer: %{time_pretransfer}\n time_redirect: %{time_redirect}\n time_starttransfer: %{time_starttransfer}\n ----------\n time_total: %{time_total}\nFaire une requête:
curl -w "@curl-format.txt" -o /dev/null -s "http://wordpress.com/"Ou sous Windows, c'est ...
curl -w "@curl-format.txt" -o NUL -s "http://wordpress.com/"
Qu'est-ce que cela fait:
-w "@curl-format.txt" dit à cURL d'utiliser notre fichier de format-o /dev/null redirige la sortie de la requête vers/dev/null-s dit à cURL de ne pas afficher de compteur de progression"http://wordpress.com/" est l'URL que nous demandons. Utilisez des guillemets, en particulier si votre URL a des paramètres de chaîne de requête "&"
Et voici ce que vous récupérez:
time_namelookup: 0.001
time_connect: 0.037
time_appconnect: 0.000
time_pretransfer: 0.037
time_redirect: 0.000
time_starttransfer: 0.092
----------
time_total: 0.164
Créer un raccourci Windows (fichier BAT)
Placez cette commande dans CURLTIME.BAT (dans le même dossier que curl.exe)
curl -w "@%~dp0curl-format.txt" -o NUL -s %*
Ensuite, vous pouvez simplement appeler ...
curltime wordpress.org
Voici la réponse:
curl -X POST -d @file server:port -w %{time_connect}:%{time_starttransfer}:%{time_total}
Toutes les variables utilisées avec -w peuvent être trouvées dans man curl.
Option 1. Pour mesurer total time:
curl -o /dev/null -s -w 'Total: %{time_total}s\n' https://www.google.com
Exemple de sortie:
Option 2. Pour obtenir time to establish connection, TTFB: time to first byte et total time:
curl -o /dev/null -s -w 'Establish Connection: %{time_connect}s\nTTFB: %{time_starttransfer}s\nTotal: %{time_total}s\n' https://www.google.com
Exemple de sortie:
Un raccourci que vous pouvez ajouter à votre fichier .bashrc, basé sur d’autres réponses ici:
function perf {
curl -o /dev/null -s -w "%{time_connect} + %{time_starttransfer} = %{time_total}\n" "$1"
}
Usage:
> perf stackoverflow.com
0.521 + 0.686 = 1.290
Ce qui suit est inspiré par la réponse de Simon. Il est autonome (ne nécessite pas de fichier de format séparé), ce qui le rend idéal pour l'inclusion dans .bashrc.
curl_time() {
curl -so /dev/null -w "\
namelookup: %{time_namelookup}s\n\
connect: %{time_connect}s\n\
appconnect: %{time_appconnect}s\n\
pretransfer: %{time_pretransfer}s\n\
redirect: %{time_redirect}s\n\
starttransfer: %{time_starttransfer}s\n\
-------------------------\n\
total: %{time_total}s\n" "$@"
}
De plus, il devrait fonctionner avec tous les arguments que curl prend normalement, puisque le "$@" ne fait que les passer. Par exemple, vous pouvez faire:
curl_time -X POST -H "Content-Type: application/json" -d '{"key": "val"}' https://postman-echo.com/post
Sortie:
namelookup: 0,125000s
connect: 0,250000s
appconnect: 0,609000s
pretransfer: 0,609000s
redirect: 0,000000s
starttransfer: 0,719000s
-------------------------
total: 0,719000s
Si vous souhaitez analyser ou résumer la latence, vous pouvez essayer Apache bench:
ab -n [number of samples] [url]
Par exemple:
ab -n 100 http://www.google.com/
Il montrera:
This is ApacheBench, Version 2.3 <$Revision: 1757674 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.Apache.org/
Benchmarking www.google.com (be patient).....done
Server Software: gws
Server Hostname: www.google.com
Server Port: 80
Document Path: /
Document Length: 12419 bytes
Concurrency Level: 1
Time taken for tests: 10.700 seconds
Complete requests: 100
Failed requests: 97
(Connect: 0, Receive: 0, Length: 97, Exceptions: 0)
Total transferred: 1331107 bytes
HTML transferred: 1268293 bytes
Requests per second: 9.35 [#/sec] (mean)
Time per request: 107.004 [ms] (mean)
Time per request: 107.004 [ms] (mean, across all concurrent requests)
Transfer rate: 121.48 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 20 22 0.8 22 26
Processing: 59 85 108.7 68 911
Waiting: 59 85 108.7 67 910
Total: 80 107 108.8 90 932
Percentage of the requests served within a certain time (ms)
50% 90
66% 91
75% 93
80% 95
90% 105
95% 111
98% 773
99% 932
100% 932 (longest request)
Une autre façon de configurer ~/.curlrc comme ceci
-w "\n\n==== cURL measurements stats ====\ntotal: %{time_total} seconds \nsize: %{size_download} bytes \ndnslookup: %{time_namelookup} seconds \nconnect: %{time_connect} seconds \nappconnect: %{time_appconnect} seconds \nredirect: %{time_redirect} seconds \npretransfer: %{time_pretransfer} seconds \nstarttransfer: %{time_starttransfer} seconds \ndownloadspeed: %{speed_download} byte/sec \nuploadspeed: %{speed_upload} byte/sec \n\n"
Donc, la sortie de curl est
❯❯ curl -I https://google.com
HTTP/2 301
location: https://www.google.com/
content-type: text/html; charset=UTF-8
date: Mon, 04 Mar 2019 08:02:43 GMT
expires: Wed, 03 Apr 2019 08:02:43 GMT
cache-control: public, max-age=2592000
server: gws
content-length: 220
x-xss-protection: 1; mode=block
x-frame-options: SAMEORIGIN
alt-svc: quic=":443"; ma=2592000; v="44,43,39"
==== cURL measurements stats ====
total: 0.211117 seconds
size: 0 bytes
dnslookup: 0.067179 seconds
connect: 0.098817 seconds
appconnect: 0.176232 seconds
redirect: 0.000000 seconds
pretransfer: 0.176438 seconds
starttransfer: 0.209634 seconds
downloadspeed: 0.000 byte/sec
uploadspeed: 0.000 byte/sec
Hey est meilleur qu'Apache Bench, a moins de problèmes avec SSL
./hey https://google.com -more
Summary:
Total: 3.0960 secs
Slowest: 1.6052 secs
Fastest: 0.4063 secs
Average: 0.6773 secs
Requests/sec: 64.5992
Response time histogram:
0.406 [1] |
0.526 [142] |∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎
0.646 [1] |
0.766 [6] |∎∎
0.886 [0] |
1.006 [0] |
1.126 [0] |
1.246 [12] |∎∎∎
1.365 [32] |∎∎∎∎∎∎∎∎∎
1.485 [5] |∎
1.605 [1] |
Latency distribution:
10% in 0.4265 secs
25% in 0.4505 secs
50% in 0.4838 secs
75% in 1.2181 secs
90% in 1.2869 secs
95% in 1.3384 secs
99% in 1.4085 secs
Details (average, fastest, slowest):
DNS+dialup: 0.1150 secs, 0.0000 secs, 0.4849 secs
DNS-lookup: 0.0032 secs, 0.0000 secs, 0.0319 secs
req write: 0.0001 secs, 0.0000 secs, 0.0007 secs
resp wait: 0.2068 secs, 0.1690 secs, 0.4906 secs
resp read: 0.0117 secs, 0.0011 secs, 0.2375 secs
Status code distribution:
[200] 200 responses
Une autre option qui est peut-être l'option la plus simple en termes de ligne de commande consiste à ajouter l'option intégrée --trace-time:
curl -X POST -d @file server:port --trace-time
Même si techniquement, il ne délivre pas les timings des différentes étapes demandés par l'OP, il affiche les horodatages des toutes étapes de la demande, comme indiqué ci-dessous. En utilisant cela, vous pouvez (assez facilement) calculer le temps que chaque étape a pris.
$ curl https://www.google.com --trace-time -v -o /dev/null
13:29:11.148734 * Rebuilt URL to: https://www.google.com/
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 013:29:11.149958 * Trying 172.217.20.36...
13:29:11.149993 * TCP_NODELAY set
13:29:11.163177 * Connected to www.google.com (172.217.20.36) port 443 (#0)
13:29:11.164768 * ALPN, offering h2
13:29:11.164804 * ALPN, offering http/1.1
13:29:11.164833 * successfully set certificate verify locations:
13:29:11.164863 * CAfile: none
CApath: /etc/ssl/certs
13:29:11.165046 } [5 bytes data]
13:29:11.165099 * (304) (OUT), TLS handshake, Client hello (1):
13:29:11.165128 } [512 bytes data]
13:29:11.189518 * (304) (IN), TLS handshake, Server hello (2):
13:29:11.189537 { [100 bytes data]
13:29:11.189628 * TLSv1.2 (IN), TLS handshake, Certificate (11):
13:29:11.189658 { [2104 bytes data]
13:29:11.190243 * TLSv1.2 (IN), TLS handshake, Server key exchange (12):
13:29:11.190277 { [115 bytes data]
13:29:11.190507 * TLSv1.2 (IN), TLS handshake, Server finished (14):
13:29:11.190539 { [4 bytes data]
13:29:11.190770 * TLSv1.2 (OUT), TLS handshake, Client key exchange (16):
13:29:11.190797 } [37 bytes data]
13:29:11.190890 * TLSv1.2 (OUT), TLS change cipher, Client hello (1):
13:29:11.190915 } [1 bytes data]
13:29:11.191023 * TLSv1.2 (OUT), TLS handshake, Finished (20):
13:29:11.191053 } [16 bytes data]
13:29:11.204324 * TLSv1.2 (IN), TLS handshake, Finished (20):
13:29:11.204358 { [16 bytes data]
13:29:11.204417 * SSL connection using TLSv1.2 / ECDHE-ECDSA-CHACHA20-POLY1305
13:29:11.204451 * ALPN, server accepted to use h2
13:29:11.204483 * Server certificate:
13:29:11.204520 * subject: C=US; ST=California; L=Mountain View; O=Google LLC; CN=www.google.com
13:29:11.204555 * start date: Oct 2 07:29:00 2018 GMT
13:29:11.204585 * expire date: Dec 25 07:29:00 2018 GMT
13:29:11.204623 * subjectAltName: Host "www.google.com" matched cert's "www.google.com"
13:29:11.204663 * issuer: C=US; O=Google Trust Services; CN=Google Internet Authority G3
13:29:11.204701 * SSL certificate verify ok.
13:29:11.204754 * Using HTTP2, server supports multi-use
13:29:11.204795 * Connection state changed (HTTP/2 confirmed)
13:29:11.204840 * Copying HTTP/2 data in stream buffer to connection buffer after upgrade: len=0
13:29:11.204881 } [5 bytes data]
13:29:11.204983 * Using Stream ID: 1 (easy handle 0x55846ef24520)
13:29:11.205034 } [5 bytes data]
13:29:11.205104 > GET / HTTP/2
13:29:11.205104 > Host: www.google.com
13:29:11.205104 > User-Agent: curl/7.61.0
13:29:11.205104 > Accept: */*
13:29:11.205104 >
13:29:11.218116 { [5 bytes data]
13:29:11.218173 * Connection state changed (MAX_CONCURRENT_STREAMS == 100)!
13:29:11.218211 } [5 bytes data]
13:29:11.251936 < HTTP/2 200
13:29:11.251962 < date: Fri, 19 Oct 2018 10:29:11 GMT
13:29:11.251998 < expires: -1
13:29:11.252046 < cache-control: private, max-age=0
13:29:11.252085 < content-type: text/html; charset=ISO-8859-1
13:29:11.252119 < p3p: CP="This is not a P3P policy! See g.co/p3phelp for more info."
13:29:11.252160 < server: gws
13:29:11.252198 < x-xss-protection: 1; mode=block
13:29:11.252228 < x-frame-options: SAMEORIGIN
13:29:11.252262 < set-cookie: 1P_JAR=2018-10-19-10; expires=Sun, 18-Nov-2018 10:29:11 GMT; path=/; domain=.google.com
13:29:11.252297 < set-cookie: NID=141=pzXxp1jrJmLwFVl9bLMPFdGCtG8ySQKxB2rlDWgerrKJeXxfdmB1HhJ1UXzX-OaFQcnR1A9LKYxi__PWMigjMBQHmI3xkU53LI_TsYRbkMNJNdxs-caQQ7fEcDGE694S; expires=Sat, 20-Apr-2019 10:29:11 GMT; path=/; domain=.google.com; HttpOnly
13:29:11.252336 < alt-svc: quic=":443"; ma=2592000; v="44,43,39,35"
13:29:11.252368 < accept-ranges: none
13:29:11.252408 < vary: Accept-Encoding
13:29:11.252438 <
13:29:11.252473 { [5 bytes data]
100 12215 0 12215 0 0 112k 0 --:--:-- --:--:-- --:--:-- 112k
13:29:11.255674 * Connection #0 to Host www.google.com left intact
J'ai créé un outil de formatage convivial pour renifler les requêtes curl afin de faciliter le débogage (voir les commentaires d'utilisation). Il contient tous les paramètres de sortie connus que vous pouvez écrire dans un format facile à lire.
https://Gist.github.com/manifestinteractive/ce8dec10dcb4725b851
Voici un Bash one-liner pour utiliser le même serveur à plusieurs reprises:
for i in {1..1000}; do curl -s -o /dev/null -w "%{time_total}\n" http://server/get_things; done
voici la chaîne que vous pouvez utiliser avec -w, contient toutes les options prises en charge par curl -w.
{"contentType":"%{content_type}","filenameEffective":"%{filename_effective}","ftpEntryPath":"%{ftp_entry_path}","httpCode":"%{http_code}","httpConnect":"%{http_connect}","httpVersion":"%{http_version}","localIp":"%{local_ip}","localPort":"%{local_port}","numConnects":"%{num_connects}","numRedirects":"%{num_redirects}","proxySslVerifyResult":"%{proxy_ssl_verify_result}","redirectUrl":"%{redirect_url}","remoteIp":"%{remote_ip}","remotePort":"%{remote_port}","scheme":"%{scheme}","size":{"download":"%{size_download}","header":"%{size_header}","request":"%{size_request}","upload":"%{size_upload}"},"speed":{"download":"%{speed_download}","upload":"%{speed_upload}"},"sslVerifyResult":"%{ssl_verify_result}","time":{"appconnect":"%{time_appconnect}","connect":"%{time_connect}","namelookup":"%{time_namelookup}","pretransfer":"%{time_pretransfer}","redirect":"%{time_redirect}","starttransfer":"%{time_starttransfer}","total":"%{time_total}"},"urlEffective":"%{url_effective}"}
sorties JSON.
curl -v --trace-time Ceci doit être fait en mode prolixe