goodrain
🐞 反馈问题：域名访问502
请先确认以下事项：
问题描述
每隔一段时间时间，访问组件的域名会突然变成502。这时候只能重启rbd-gateway，才能正常访问。
该问题是否可以稳定重现？
可重现
重现步骤
我查看了下监控，每次rbd-gateway组件的内存使用到差不多900M的时候，就会出现这个问题。
截图
日志
期望结果
希望能尽快解决
解决方案（可选）
No response
操作系统 && Rainbond 版本
Alibaba Cloud Linux 3、1.28.15-aliyun.1、v6.2.0-release-842bfbd-2025-04-03-11
是否愿意提交 PR 解决该问题？
- [ ] 我愿意提交 PR 来解决该问题
Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑🤝🧑👫🧑🏿🤝🧑🏻👩🏾🤝👨🏿👬🏿
Title: 🐞 Feedback question: Domain name access 502
Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑🤝🧑👫🧑🏿🤝🧑🏻👩🏾🤝👨🏿👬🏿
@Lanka
能否提供下当前 rbd-gateway 的镜像，可以通过如下方式获取
kubectl get po -nrbd-system -l name=rbd-gateway -oyaml | grep image:
Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑🤝🧑👫🧑🏿🤝🧑🏻👩🏾🤝👨🏿👬🏿
Can you provide the current rbd-gateway image? You can obtain it in the following way
kubectl get po -nrbd-system -l name=rbd-gateway -oyaml | grep image:
能否提供下当前 rbd-gateway 的镜像，可以通过如下方式获取
kubectl get po -nrbd-system -l name=rbd-gateway -oyaml | grep image:
刚刚看到你们发布新版本，直接升级了。目前是这个:
刚刚看了下，升级前就是502，升级后还是502。然后只能重启rbd-gateway恢复
Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑🤝🧑👫🧑🏿🤝🧑🏻👩🏾🤝👨🏿👬🏿
Can you provide the current rbd-gateway image? You can obtain it in the following way
kubectl get po -nrbd-system -l name=rbd-gateway -oyaml | grep image:
I just saw you releasing a new version and upgraded it directly. It's currently this:
组件在发生502的时候是否有过构建、滚动更新、开关对外服务等操作？rbd-gateway 自身是否发生过异常重启。当前可能有以下已知问题：
-
如果组件重启或滚动更新，ingress-controller 由于未正确更新组件后端的访问地址，导致无法访问。这个情况在 rbd-api 服务升级后，重新开关组件对外服务即可解决。
-
gateway 服务的 ingress-controller 由于并发写入问题，导致重启，这个也许你可以先尝试更换以下镜像，观察一段时间
registry.cn-hangzhou.aliyuncs.com/goodrain/apisix-ingress-controller:011502
如果还有问题，提供下 rbd-gateway 这个 pod 中的两个容器的日志。
Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑🤝🧑👫🧑🏿🤝🧑🏻👩🏾🤝👨🏿👬🏿
Have the components undergone construction, rolling updates, switch external services, etc. when 502 occurs? Whether an abnormal restart occurred in rbd-gateway itself. There may be the following known issues:
-
If the component restarts or rolls updating, the ingress-controller fails to access the access address of the component backend because it does not correctly update the access address of the component backend. This situation can be solved after the rbd-api service is upgraded, and the external service of the reswitch component can be solved.
-
The gateway service's ingress-controller caused by concurrent write problems, which caused a restart. Maybe you can try to replace the following image first and observe for a period of time.
registry.cn-hangzhou.aliyuncs.com/goodrain/apisix-ingress-controller:011502
If there are still problems, provide the logs of the two containers in the rbd-gateway pod.