baijinge0310

1. new一个页面时会设置一个隐藏token 2. AddToken时会往接收到请求的pod你内存中添加，而此时其他pod的内存中不会有 3. 在页面的curd操作时，会触发CheckToken，如果恰好又发送到了同一个pod，那么可以返回True，如果分发到其他内存中没有这个token的pod，则会返回False导致鉴权失败 4. 一些CheckToken触发的位置

I got the same problem

[dashboard.txt](https://github.com/milvus-io/milvus/files/11767231/dashboard.txt) You can try this. You need to add a job to the configuration file `prometheus.yml` of the prometheus service, and fill in the metric address of milvus.

> sounds like a cgo known issue. /assign @xige-16 /unassign I released and loaded the collection again and again, and gradually stabilized. After each load, there was an extra 2.2...

There is the latest progress. After the collection of the load+release sq8 index type, the memory residue caused by it can be completely cleaned up through the collection of the...

This is a single-node service started by docker-compose. Since the log size exceeds the github limit, I have sent the standalone log to Mr. Xiaolong of “Milvus用户交流群102”.

> @baijinge0310 Could you please refer this [script](https://github.com/milvus-io/milvus/tree/master/deployments/export-log) to export the whole Milvus logs for investigation? > > /assign @baijinge0310 https://pan.baidu.com/s/1FPTaYYs6Zm9Aw7Hdql3LQg?pwd=qvqr I uploaded the log to Baidu online disk.I started...

> > > > 啥是H5渠道 就是网页上这种，html5