renatomefi / php-fpm-healthcheck Goto Github PK

View Code? Open in Web Editor NEW

476.0 17.0 53.0 98 KB

A POSIX compliant sh script to healthcheck PHP fpm status, can be used only for pinging or check for specific metrics

License: MIT License

Makefile 9.43% Shell 38.19% Python 52.37%

docker php php-fpm health-check kubernetes livenessprobe

php-fpm-healthcheck's People

Contributors

Stargazers

Watchers

Forkers

eltonoliveira rdohms mrnomnom kuno1 0x00031337 itshosted smatyas ansilh roynasser ahorcholle yaowenqiang run100 ijenkac matterstech hosseinsalemi broekhuisholding lexxksb jelovac anare royletron peak fernandesgabriel empty-set brunopadz anwarmuhamat wandersonwhcr andrew-demb cluster2a blodan finsecsystems wjts wladimir-n eduartua trusted-rocks mpcabanto cn27001 wallacetan package-health daniel-lizzander adrias-online kahu-app legalease mnpenner noone-silent simplew23 bitjip alifattahi mannew oleg-kolzhanov vgeorgio christophbrejla blins tkaefer

php-fpm-healthcheck's Issues

Document development, how to contribute

Context

Document on how to contribute to this project

Tips

Basically run make qa

Missing implementation of exit status codes

Hi,

When command cgi-fcgi fails to connect to a php-fpm instance the script exists.

Shouldn't there be an implementation for the following exit codes as described in the script header:

2,9,111 - Couldn't connect to PHP fpm, is it running?

For example:

    if test "$FPM_STATUS" = "Could not connect to $1"; then
        >&2 printf "Failed to connect to php-fpm instace via $1. Is it running? \\n";
        exit 2;
    fi;

Ubuntu - cgi-fcgi isn't terminating

Hi all,

When trying to use this script with ubuntu I'm seeing an ever-growing number of active cgi-fcgi instances and the health checks aren't terminating. Any idea why?

Test failed metrics

Context

Test when a metrics has fail to success

Suggestion

Use the metric --accepted-conn since it increases on every time a request is served by php-fpm
Also test with multiple metrics in different orders

Script Bugs with Image based on Debian 9 stretch

Environment:

# cat /etc/os-release 
PRETTY_NAME="Debian GNU/Linux 9 (stretch)"
NAME="Debian GNU/Linux"
VERSION_ID="9"
VERSION="9 (stretch)"
ID=debian
HOME_URL="https://www.debian.org/"
SUPPORT_URL="https://www.debian.org/support"
BUG_REPORT_URL="https://bugs.debian.org/

How to reproduce :

Run php-fpm-healthcheck --listen-queue=10
And you will get this error :

/usr/local/bin/php-fpm-healthcheck: 78: test: Illegal number:

Readiness probe error not handled due to php-fpm-healthcheck timeout

When php-fpm stuck and didn't respond, php-fpm-healthckeck can't handle this, and timeouts to connect to the port.

root@deployment-php-5bddfd7964-7ctmf:/app# php-fpm-healthcheck -v Trying to connect to php-fpm via: localhost:9000/status ^C
So, php-fpm-healthcheck doesn't return fail status in a reasonable time.

Since K8s doesn't handle timeouts healthcheck, it doesn't work. Because Kubernetes remove the pod from service only when heathcheck returning fail event.
Readiness probe errored: rpc error: code = DeadlineExceeded desc = context deadline exceeded
Readiness probe errored, but pod didn't remove from services.

Is there any way to setup a reasonable timeout to connect to the PHP port for the php-fpm-healthcheck?

Support all php-fpm metrics

Context

Support all php-fpm metrics which can be seen via: php-fpm-healthcheck -v
Of course ignoring non numeric/measurable data

Impact

Documentation and tests

36: set: Illegal option -o pipefail

Installed as

wget -O /usr/local/bin/php-fpm-healthcheck \
https://raw.githubusercontent.com/renatomefi/php-fpm-healthcheck/master/php-fpm-healthcheck \
&& chmod +x /usr/local/bin/php-fpm-healthcheck

result is:

# php-fpm-healthcheck
/usr/local/bin/php-fpm-healthcheck: 36: set: Illegal option -o pipefail

# php-fpm-healthcheck -v
/usr/local/bin/php-fpm-healthcheck: 36: set: Illegal option -o pipefail

I can't understand the livenessProbe example, Can you explain it in detail?

https://github.com/renatomefi/php-fpm-healthcheck#livenessprobe

--listen-queue=10 # fails if there are more than 10 processes waiting in the fpm queue
In this case, pod should be scaleUp, rather than reset？

--accepted-conn=5000 # fails after fpm has served more than 5k requests, this will force the pod to reset, use with caution
In this case, pod don't have to reset., Why do you configure it this way?

So how do you configure it in production？

Test other distros than alpine

Context

Currently the test matrix only tests alpine images, would be good to be able to test strecht for instance.

Impact

tests might differ, I recommend using pytest mark feature to separate those distro specific
docker.sh has to adhere to the marks, build and run this matrix, maybe the make file could inform which Dockerfile to look at

centos not support usage of ' | tail +5'

I use this script on centos, but failed.

error output like this:

[root@root /] echo "any text" | tail +5
[root@root /] tail: cannot open `+5' for reading: No such file or directory

[root@root /] FCGI_CONNECT=/dev/shm/php-fpm.socket FCGI_STATUS_PATH=/fpm/status ./php-fpm-healthcheck.sh --verbose
[root@root /] Trying to connect to php-fpm via: /dev/shm/php7-fpm.socket/fpm/status
[root@root /] tail: cannot open `+5' for reading: No such file or directory

Metric check for listen queue len seems wrong

Based on https://serverfault.com/a/355546 the listen queue len value is decreased on connection. So initially the value might be for example 128. When I run the check via php-fpm-healthcheck --listen-queue-len=10 it fails with 'listen queue len' value '128' is greater than expected '10'. In this case this should only fail if the value of listen queue len is lower than 10.

Understanding FastCGI (with Ubuntu 22.04)

⏰ Background

I am building "PHP-FPM + NGINX" and "PHP + Apache2" images with S6 Overlay 3.1.2.1

👉 Problem I am trying to solve

Some users are reporting that NGINX is starting before PHP-FPM: https://community.serversideup.net/t/502-bad-gateway-on-google-cloud-run/700/3

🤔 My Approach to a Solution

I found this lovely project, but I am confused on what FastCGI really is and why it's a dependency for checking the status of PHP-FPM.

What I am attempting to do is:

Start services in order, but ensure each service is healthy before continuing to the next

Here is how I am starting my services:

graph TD;
    A[S6 Overlay] --> B[PHP-FPM];
     B[PHP-FPM] --> C[NGINX];

The folks over at S6 Overlay have been very helpful and pointed me in the right direction for the script side of things: just-containers/s6-overlay#482

Now I just need to get that "definition of healthy" with an exit code of 0.

That's when I found this project and saw the dependency for FastCGI.

I googled "Install FastCGI Ubuntu 22.04" and this was the first result: https://installati.one/ubuntu/22.04/fcgiwrap/

👉 My Problem

Even with having fcgiwrap installed, I am getting this error:

Error:

Make sure fcgi is installed (i.e. apk add --no-cache fcgi). Aborting.

👨‍💻 My Source Code

I have all my source code available on GitHub: https://github.com/serversideup/docker-php/tree/feature/add-readiness/src

Note: The fpm-nginx image is built from the fpm image (so you understand the image dependency and what's included).

❓ Questions

Is there a lightweight way to run this FastCGI server on Ubuntu 22.04 without it installing a bunch of other things in my Docker Image?

Thanks for all your hard work! This project is incredible!! 🤩

Add tests for 7.4-fpm-buster

Add a new Dockerfile for buster:
Add it to the Makefile in /test/Dockerfile-buster
Add to the circleci build

CircleCI Error

Currently our master branch has a CircleCI failed run.

https://app.circleci.com/pipelines/github/renatomefi/php-fpm-healthcheck/94/workflows/50a8efa6-4974-4aca-a874-b924098b15cd/jobs/457

Permission denied (publickey).
fatal: Could not read from remote repository.

Please make sure you have the correct access rights
and the repository exists.

exit status 128

CircleCI received exit code 128

@renatomefi will we continue using CircleCI?

When environment variables exceeds 8kb fpm will close the connection

Context

If you have more than 8kb exported in your environment, they'll will be send by default towards the fcgi call to php fpm, this makes fpm close the connection causing either exit status code of 104 or 253depending on when it happens in the internal calls.

How to fix

Send only the needed variables towards php fpm

It could be possible by replacing the fcgi call with:
env -i cgi-fcgi...

env -i meant to clear the current variables in the session

Log

An strace would look like:

strace cgi-fcgi -bind -connect /var/run/php-fpm.sock
execve("/usr/bin/cgi-fcgi", ["cgi-fcgi", "-bind", "-connect", "/var/run/php-fpm.sock"], 0x7ffed28a0c08 /* 191 vars */) = 0
arch_prctl(ARCH_SET_FS, 0x7f1e94c06b88) = 0
set_tid_address(0x7f1e94c06bc0)         = 16859
open("/etc/ld-musl-x86_64.path", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
open("/lib/libfcgi.so.0", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
open("/usr/local/lib/libfcgi.so.0", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
open("/usr/lib/libfcgi.so.0", O_RDONLY|O_CLOEXEC) = 3
fcntl(3, F_SETFD, FD_CLOEXEC)           = 0
fstat(3, {st_mode=S_IFREG|0755, st_size=38488, ...}) = 0
read(3, "\177ELF\2\1\1\0\0\0\0\0\0\0\0\0\3\0>\0\1\0\0\0\240+\0\0\0\0\0\0"..., 960) = 960
mmap(NULL, 2138112, PROT_READ|PROT_EXEC, MAP_PRIVATE, 3, 0) = 0x7f1e94771000
mmap(0x7f1e94979000, 8192, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED, 3, 0x8000) = 0x7f1e94979000
close(3)                                = 0
mprotect(0x7f1e94979000, 4096, PROT_READ) = 0
mprotect(0x7f1e94c03000, 4096, PROT_READ) = 0
mprotect(0x561b68d55000, 4096, PROT_READ) = 0
rt_sigaction(SIGPIPE, NULL, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=0}, 8) = 0
rt_sigprocmask(SIG_UNBLOCK, [RT_1 RT_2], NULL, 8) = 0
rt_sigaction(SIGPIPE, {sa_handler=0x7f1e9477694e, sa_mask=[], sa_flags=SA_RESTORER, sa_restorer=0x7f1e949c10ea}, NULL, 8) = 0
rt_sigaction(SIGUSR1, NULL, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=0}, 8) = 0
rt_sigaction(SIGUSR1, {sa_handler=0x7f1e94776af8, sa_mask=[], sa_flags=SA_RESTORER, sa_restorer=0x7f1e949c10ea}, NULL, 8) = 0
socket(AF_UNIX, SOCK_STREAM, 0)         = 3
connect(3, {sa_family=AF_UNIX, sun_path="/var/run/php-fpm.sock"}, 23) = 0
write(3, "\1\1\0\1\0\10\0\0\0\1\0\0\0\0\0\0", 16) = 16
brk(NULL)                               = 0x561b693c7000
brk(0x561b693ca000)                     = 0x561b693ca000
write(3, "\1\4\0\1\37\370\0\0008\16FEEDBACK_COLLECTOR_HTT"..., 8192) = 8192
write(3, "\1\4\0\1\1\245\3\0SERVICE_PORT8081\30\3BOTKIT"..., 440) = -1 EPIPE (Broken pipe)
--- SIGPIPE {si_signo=SIGPIPE, si_code=SI_USER, si_pid=16859, si_uid=0} ---
rt_sigreturn({mask=[]})                 = -1 EPIPE (Broken pipe)
fcntl(3, F_GETFL)                       = 0x2 (flags O_RDWR)
fcntl(3, F_SETFL, O_RDWR|O_NONBLOCK)    = 0
select(4, [3], [3], NULL, NULL)         = 2 (in [3], out [3])
read(3, "", 8192)                       = 0
shutdown(3, SHUT_WR)                    = 0
poll([{fd=3, events=POLLIN}], 1, 2000)  = 1 ([{fd=3, revents=POLLIN|POLLHUP}])
read(3, "", 1024)                       = 0
close(3)                                = 0
exit_group(-3)                          = ?
+++ exited with 253 +++
/opt/search-api # strace cgi-fcgi -bind -connect /var/run/php-fpm.sock
execve("/usr/bin/cgi-fcgi", ["cgi-fcgi", "-bind", "-connect", "/var/run/php-fpm.sock"], 0x7ffdc1d0c428 /* 191 vars */) = 0
arch_prctl(ARCH_SET_FS, 0x7feba6734b88) = 0
set_tid_address(0x7feba6734bc0)         = 16862
open("/etc/ld-musl-x86_64.path", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
open("/lib/libfcgi.so.0", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
open("/usr/local/lib/libfcgi.so.0", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
open("/usr/lib/libfcgi.so.0", O_RDONLY|O_CLOEXEC) = 3
fcntl(3, F_SETFD, FD_CLOEXEC)           = 0
fstat(3, {st_mode=S_IFREG|0755, st_size=38488, ...}) = 0
read(3, "\177ELF\2\1\1\0\0\0\0\0\0\0\0\0\3\0>\0\1\0\0\0\240+\0\0\0\0\0\0"..., 960) = 960
mmap(NULL, 2138112, PROT_READ|PROT_EXEC, MAP_PRIVATE, 3, 0) = 0x7feba629f000
mmap(0x7feba64a7000, 8192, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED, 3, 0x8000) = 0x7feba64a7000
close(3)                                = 0
mprotect(0x7feba64a7000, 4096, PROT_READ) = 0
mprotect(0x7feba6731000, 4096, PROT_READ) = 0
mprotect(0x55cb34157000, 4096, PROT_READ) = 0
rt_sigaction(SIGPIPE, NULL, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=0}, 8) = 0
rt_sigprocmask(SIG_UNBLOCK, [RT_1 RT_2], NULL, 8) = 0
rt_sigaction(SIGPIPE, {sa_handler=0x7feba62a494e, sa_mask=[], sa_flags=SA_RESTORER, sa_restorer=0x7feba64ef0ea}, NULL, 8) = 0
rt_sigaction(SIGUSR1, NULL, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=0}, 8) = 0
rt_sigaction(SIGUSR1, {sa_handler=0x7feba62a4af8, sa_mask=[], sa_flags=SA_RESTORER, sa_restorer=0x7feba64ef0ea}, NULL, 8) = 0
socket(AF_UNIX, SOCK_STREAM, 0)         = 3
connect(3, {sa_family=AF_UNIX, sun_path="/var/run/php-fpm.sock"}, 23) = 0
write(3, "\1\1\0\1\0\10\0\0\0\1\0\0\0\0\0\0", 16) = 16
brk(NULL)                               = 0x55cb34ba0000
brk(0x55cb34ba3000)                     = 0x55cb34ba3000
write(3, "\1\4\0\1\37\370\0\0008\16FEEDBACK_COLLECTOR_HTT"..., 8192) = 8192
write(3, "\1\4\0\1\1\245\3\0SERVICE_PORT8081\30\3BOTKIT"..., 440) = -1 EPIPE (Broken pipe)
--- SIGPIPE {si_signo=SIGPIPE, si_code=SI_USER, si_pid=16862, si_uid=0} ---
rt_sigreturn({mask=[]})                 = -1 EPIPE (Broken pipe)
fcntl(3, F_GETFL)                       = 0x2 (flags O_RDWR)
fcntl(3, F_SETFL, O_RDWR|O_NONBLOCK)    = 0
select(4, [3], [3], NULL, NULL)         = 2 (in [3], out [3])
read(3, "", 8192)                       = 0
shutdown(3, SHUT_WR)                    = 0
poll([{fd=3, events=POLLIN}], 1, 2000)  = 1 ([{fd=3, revents=POLLIN|POLLHUP}])
read(3, "", 1024)                       = 0
close(3)                                = 0
exit_group(-3)                          = ?
+++ exited with 253 +++

Document and test Docker healthcheck

Context

Docker has its internal healthcheck https://docs.docker.com/engine/reference/builder/#healthcheck
Would be good to document and test it

Impact

It might require to split the test process since you probably have to use docker inspect to get the current healthcheck status

Disable access logs

Is there a way to disable access logs for path /status in php-fpm?

21/Oct/2019:19:58:00 +0000 "GET /status" 200

I run this in Kubernetes and have separate containers for php-fpm and nginx. I don's see the logs from the line above in nginx logs, just in php-fpm logs.

not working in debian linux with FPM socket communication

Hi guys,
We had to switch from alpine to debian linux (unfortunately). Since then, the healthcheck script is no longer running. I only receive the answer: Make sure fcgi is installed (i.e. apk add --no-cache fcgi). Aborting.
I'm running the script with: FCGI_CONNECT=/var/run/socket/php5-fpm.sock php-fpm-healthcheck

Could we find out the reason for it?

Test fcgi socket connection

Context

Currently we're only testing when sockets are running on localhost:9000, create a test where we connect via socket for instance
How to change FCGI connection is described here: https://github.com/renatomefi/php-fpm-healthcheck#connection-via-socket-or-another-host

Impact

The current docker image would have to be either altered in runtime or build a new one with new php-fpm pool configuration to create a socket

renatomefi / php-fpm-healthcheck Goto Github PK

php-fpm-healthcheck's People

Contributors

Stargazers

Watchers

Forkers

php-fpm-healthcheck's Issues

Context

Tips

Context

Suggestion

Environment:

How to reproduce :

Context

Impact

Context

Impact

⏰ Background

👉 Problem I am trying to solve

🤔 My Approach to a Solution

👉 My Problem

👨‍💻 My Source Code

❓ Questions

Context

How to fix

Log

Context

Impact

Context

Impact

Recommend Projects

Recommend Topics

Recommend Org