Skip to the content.

PHP 8.0 现在不支持吗?

No description provided.

Test Phalcon integration with TARS

No description provided.

已经做的修改:

fork了一份代码,https://github.com/luoxiaojun1992/tars-php-phalcon 修复了servers.php配置,因为支持了多servant,原作者的配置文件会报错 删除了原作者的一些测试代码(格式比较乱),增加了http、tars的示例接口 对目录结构适当做了一些调整 更新文档 用tarsphp的docker镜像搭建的平台,在node容器安装了phalcon扩展,测试了http和tars可以调用

遇到的问题:

第一次部署phalcon服务,平台弹窗返回成功,之后都是返回失败,实际上服务已经更新重启,平台上显示的服务状态也是active。Laravel和Lumen的服务部署也是同样的问题,可能是docker镜像平台的问题。

子任务ID 应用 服务 节点 命令 开始时间 结束时间 状态 执行信息

d128c8f5d1bc4610b6ba5b256d3a852e PHPTest LaravelTars 172.18.0.4 patch_tars 2020-04-22 22:55:59 2020-04-22 22:56:07 EM_I_FAILED [ServantProxy::invoke timeout:3000,servant:tars.tarsAdminRegistry.AdminRegObj,func:restartServer,adaptertcp -h 172.18.0.3 -p 12000,reqid:22]

子任务ID 应用 服务 节点 命令 开始时间 结束时间 状态 执行信息

216605b3a2df4c1385e023627071d747 PHPTest TarsPhalcon 172.18.0.4 patch_tars 2020-04-22 22:57:10 2020-04-22 22:57:14 EM_I_FAILED [ServantProxy::invoke timeout:3000,servant:tars.tarsAdminRegistry.AdminRegObj,func:restartServer,adaptertcp -h 172.18.0.3 -p 12000,reqid:30]

7645a1714f80421d8510b65f1f79942e PHPTest TarsPhalcon 172.18.0.3 patch_tars 2020-04-22 22:57:14 2020-04-22 22:57:14 EM_I_CANCEL

还需要研究完善的:

支持phalcon的路由(如果有的话),目前是采用的Tars\Command的默认路由,将tars的请求响应上下文转换为phalcon的请求响应上下文 (Done) 拉取tars的平台配置合并到phalcon的项目配置中 phalcon的log适配集成tarsLog 参考其他的phalcon集成swoole的项目做一些优化(比如防止内存泄露等)

Modify PHP-Zipkin library to use swoole async report

No description provided.

Demo for distributed tracing

No description provided.

tarsphp+swoft project start failed on Tars web. I manually start the service, but it can’t be reached in browser

I did a telnet on the node server. Port 18306 is connectable Here’s the config content of my service.

enableset=n setdivision=NULL node=tars.tarsnode.ServerObj@tcp -h 172.25.0.5 -p 19386 -t 60000 deactivating-timeout=3000 app=PHPTest server=swoft localip=172.25.0.5 local=tcp -h 127.0.0.1 -p 9100 -t 3000 basepath=/usr/local/app/tars/tarsnode/data/PHPTest.swoft/bin/ datapath=/usr/local/app/tars/tarsnode/data/PHPTest.swoft/data/ logpath=/usr/local/app/tars/app_log/ loglevel=DEBUG logsize=15M log=tars.tarslog.LogObj config=tars.tarsconfig.ConfigObj notify=tars.tarsnotify.NotifyObj logLevel=DEBUG php=/usr/bin/php buffer_output_size=12582912 open_tcp_nodelay=1 open_eof_check=0 open_eof_split=0 task_worker_num=1 dispatch_mode=2 daemonize=1 allow endpoint=tcp -h 172.25.0.5 -p 9100 -t 60000 handlegroup=PHPTest.swoft.objAdapter maxconns=200000 protocol=not_tars queuecap=10000 queuetimeout=60000 servant=PHPTest.swoft.obj threads=5 asyncthread=12 locator=tars.tarsregistry.QueryObj@tcp -h 172.25.0.3 -p 17890 sync-invoke-timeout=20000 async-invoke-timeout=20000 refresh-endpoint-interval=60000 stat=tars.tarsstat.StatObj property=tars.tarsproperty.PropertyObj report-interval=60000 modulename=PHPTest.swoft sample-rate=100000 max-sample-count=50

Please check the node tars.application.server.local in your configure file. Your service is on 9100 port.

Investigate Github actions

No description provided.

怎么没有phalcon 整合的文档呢?

名字也是错的!!!

Phalcon我们还在接入,敬请期待。

您可以先参考这位开发者的代码,文档我们近期会完善好,感谢您的支持。 https://github.com/iamloso/tars-php-phalcon

tars监控swoole进程导致服务器内部502

测试重现步骤: max_request设置为5 (实际生产环境配置为1000000) worker接收到http最大请求次数超过5,首先swoole机制原因worker进程会自动重启, 推断:tars监听到的woker进程有变化,并将其kill掉再启动一次? 附swoole官方文档解释 https://wiki.swoole.com/wiki/page/p-max_request.html 附swoole.log中的记录: [2019-04-18 09:33:32 #2357.4] NOTICE Server is shutdown now. [2019-04-18 09:33:32 $2358.0] ERROR swProcessPool_shutdown(:339): waitpid(2369) failed. Error: No child processes[10]. [2019-04-18 09:33:32 $2358.0] ERROR swProcessPool_shutdown(:339): waitpid(2370) failed. Error: No child processes[10]. [2019-04-18 09:33:32 $2358.0] ERROR swProcessPool_shutdown(:339): waitpid(2371) failed. Error: No child processes[10]. [2019-04-18 09:33:32 $2358.0] ERROR swProcessPool_shutdown(:339): waitpid(2372) failed. Error: No child processes[10]. [2019-04-18 09:33:32 $2358.0] ERROR swProcessPool_shutdown(:339): waitpid(2373) failed. Error: No child processes[10]. [2019-04-18 09:33:32 $2358.0] ERROR swProcessPool_shutdown(:339): waitpid(2374) failed. Error: No child processes[10]. [2019-04-18 09:33:32 $2358.0] ERROR swProcessPool_shutdown(:339): waitpid(2375) failed. Error: No child processes[10]. [2019-04-18 09:33:32 $2358.0] ERROR swProcessPool_shutdown(:339): waitpid(2376) failed. Error: No child processes[10]. [2019-04-18 09:33:32 $2358.0] ERROR swProcessPool_shutdown(:339): waitpid(2377) failed. Error: No child processes[10]. [2019-04-18 09:33:32 $2358.0] ERROR swProcessPool_shutdown(:339): waitpid(2378) failed. Error: No child processes[10]. [2019-04-18 09:33:32 $2358.0] ERROR swManager_loop(:474): waitpid(2379) failed. Error: No child processes[10]. [2019-04-18 09:33:32 $2358.0] ERROR swManager_loop(:474): waitpid(2380) failed. Error: No child processes[10]. [2019-04-18 09:33:32 $2358.0] ERROR swManager_loop(:474): waitpid(2381) failed. Error: No child processes[10]. [2019-04-18 09:33:32 $2358.0] ERROR swManager_loop(:474): waitpid(2382) failed. Error: No child processes[10]. [2019-04-18 09:33:32 $2358.0] ERROR swManager_kill_user_worker(:609): waitpid(2383) failed. Error: No child processes[10]. [2019-04-18 09:33:32 $2358.0] ERROR swManager_kill_user_worker(:609): waitpid(2384) failed. Error: No child processes[10]. [2019-04-18 09:33:32 #2357.4] ERROR swFactoryProcess_shutdown(:62): waitpid(2358) failed. Error: No child processes[10]. Server has been started. (master PID: 2415, manager PID: 2416)

tars的确会对swoole进程进行监控,使用max_request参数的原因是?

这个可能呢是监听事件的问题, tars 拉起来的 swoole ,worker 退出后,master 不能获得信号量,导致master 不能拉起退出的worker

问下啊 这个问题解决了么,用max request NUM主要是为了内存泄漏

tars的确会对swoole进程进行监控,使用max_request参数的原因是?