Open TreehouseFalcon opened 1 year ago
The statistics in the dashboard is not real-time. They are updated only after dataset pulled from the database.
同样的问题,那什么时候或者多久会触发数据库的更新呢
Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑🤝🧑👫🧑🏿🤝🧑🏻👩🏾🤝👨🏿👬🏿
The same problem, when or how long will the update of the database be triggered?
The statistics in the dashboard is not real-time. They are updated only after dataset pulled from the database.
How often does this occur? It seems to take well over an hour even with feedback and items going in and out. Does it only happen on master application startup? If so it seems like that would be reported as unintended behavior quite frequently.
仪表板中的统计数据不是实时的。它们仅在从数据库中提取数据集后才会更新。
这种情况多久发生一次?即使有反馈和物品进出,似乎也需要一个多小时。它只发生在主应用程序启动时吗?如果是这样,那么这似乎会经常被报告为意外行为。
我的feedback利用接口插入之后,通常需要隔天才能在dashboard上显示。对一个用户插入feedback之后,针对该用户用接口查询该用户的所有反馈也是查不到的。需要隔天才会出现,我已经将模型训练的时间设置的很短,但还是需要隔天才能出现。请问加载feedback的时间间隔是怎么控制的呢?
Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑🤝🧑👫🧑🏿🤝🧑🏻👩🏾🤝👨🏿👬🏿
The stats in the dashboard are not live. They are only updated after the dataset is fetched from the database.
How often does this happen? Even with feedback and items coming in and out, it seems to take over an hour. Does it only happen when the main application starts? If so, then this seems to be frequently reported as unexpected behavior.
After my feedback is inserted using the interface, it usually takes the next day to be displayed on the dashboard. After inserting feedback for a user, it is impossible to find all the user's feedback by using the interface for the user. It takes the next day to appear. I have set the model training time to be very short, but it still takes the next day to appear. How to control the time interval of loading feedback?
仪表板中的统计数据不是实时的。它们仅在从数据库中提取数据集后才会更新。
这种情况多久发生一次?即使有反馈和物品进出,似乎也需要一个多小时。它只发生在主应用程序启动时吗?如果是这样,那么这似乎会经常被报告为意外行为。
我的feedback利用接口插入之后,通常需要隔天才能在dashboard上显示。对一个用户插入feedback之后,针对该用户用接口查询该用户的所有反馈也是查不到的。需要隔天才会出现,我已经将模型训练的时间设置的很短,但还是需要隔天才能出现。请问加载feedback的时间间隔是怎么控制的呢?
如果你位于中国,那么大概率是时区问题。。试着把数据库里的time_stamp减去8小时,再重启gorse,你会惊奇地发现它更新了。。我为了找这个原因找了一晚上orz我的处理办法是直接在添加反馈条目前,就把time_stamp减去8小时;你可以尝试修改本地时区/docker时区,反正我没搞成功,改了之后甚至超前了8小时,真是醉了,干脆存之前就减去8小时了
Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑🤝🧑👫🧑🏿🤝🧑🏻👩🏾🤝👨🏿👬🏿
Statistics in the dashboard are not real-time. They are only updated after the dataset is fetched from the database.
How often does this happen? Even with feedback and items coming in and out, it seemed to take over an hour. Does it only happen when the main application starts? If so, then this seems to be frequently reported as unexpected behavior.
After my feedback is inserted using the interface, it usually takes the next day to be displayed on the dashboard. After inserting feedback for a user, even if you use the interface to query all the feedback of the user, you will not be able to find it. It takes the next day to appear. I have set the model training time to be very short, but it still takes the next day to appear. How is the time interval for loading feedback controlled? ![image](https://private-user-images.githubusercontent.com/61278806/261474116-d4a0d888-8cf8-4236-b48c-5fb68e57c201.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJ naXRodWIuY29tIiwiYXVkIjoicmF3LmdpdGh1YnVzZXJjb250ZW50LmNvbSIsImtleSI6ImtleTUiLCJleHAiOjE3MDUxNjczMzYsIm5iZiI6MTcwNTE2NzAzNiwicGF0aCI6Ii82MTI3ODgwNi8yNjE 0NzQxMTYtZDRhMGQ4ODgtOGNmOC00MjM2LWI0OGMtNWZiNjhlNTdjMjAxLnBuZz9YLUFtei1BbGdvcml0aG09QVdTNC1ITUFDLVNIQTI1NiZYLUFtei1DcmVkZW50aWFsPUFLSUFWQ09EWUxTQTUzUFFLNFpBJ TJGMjAyNDAxMTMlMkZ1cy1lYXN0LTElMkZzMyUyRmF3czRfcmVxdWVzdCZYLUFtei1EYXRlPTIwMjQwMTEzVDE3MzAzNlomWC1BbXotRXhwaXJlcz0zMDAmWC1BbXotU2lnbmF0dXJlPTU2ZjA4ZTJjY WQwNmU3OTQzYzAwMDQwOWMwMTAyMDE0YWZmZWY2ZTRjNWU3ZTNkY2MwZDIxMjg0OTg5MWNhNjUmWC1BbXotU2lnbmVkSGVhZGVycz1ob3N0JmFjdG9yX2lkPTAma2V5X2lkPTAmcmVwb19pZD0w In0.8Ww7Vhe5R6H1ZlkyoDn_aZNpvIcUgf8Vv_SP9mpw4sk)
If you are located in China, then it is most likely a time zone issue. . Try subtracting 8 hours from the time_stamp in the database and restart gorse. You will be surprised to find that it is updated. . I searched orz all night to find this reason. My solution was to subtract 8 hours from time_stamp before adding the feedback item; you can try to modify the local time zone/docker time zone. Anyway, I didn't succeed. After changing it, even It's 8 hours ahead. It's really drunk. I might as well subtract 8 hours before saving.
Gorse version v0.4.14 (gorse-in-one, per dashboard master version)
Describe the bug The web dashboard only shows accurate metrics on startup. I've had it running for hours a couple of times but nothing changes.
To Reproduce Simply start Gorse and insert some items/positively marked feedback. The 5 metrics on the overview page do not change.
Expected behavior I expect the dashboard to be relatively real-time. I'm not sure what the metrics refresh period is, but it seemed to be stale for well over an hour.
Additional context Is this period a configuration option I should be changing somewhere?