使用HTTP POST将OpenTSDB写入Bigtable不起作用(使用Kubernetes( [英] Writing OpenTSDB to Bigtable with HTTP POST not working (using Kubernetes(

问题描述

更新了更多信息

Updated with more information

我正在按照此指南在Bigtable上设置OpenTSDB: https://cloud.google.com/solutions/opentsdb-cloud-platform

I am trying to set up OpenTSDB on Bigtable, following this guide: https://cloud.google.com/solutions/opentsdb-cloud-platform

效果很好，一切都很好.

Works well, all good.

现在，我正尝试使用LoadBalancer(类型)打开opentsdb-write服务.看起来也很好.

Now I was trying to open the opentsdb-write service with a LoadBalancer (type). Seems to work well, too.

注意:使用GCP负载平衡器.

Note: using a GCP load balancer.

然后，我正在使用失眠将POST发送到./api/put端点-并按预期方式获得了204(此外，使用?details也没有显示错误，?sync也没有显示错误)(请参见 http://opentsdb.net/docs/build/html/api_http/put.html )

I am then using insomnia to send a POST to the ./api/put endpoint - and I get a 204 as expected (also, using the ?details shows no errors, neither does the ?sync) (see http://opentsdb.net/docs/build/html/api_http/put.html)

查询数据时(在./api/query上获取)，我看不到数据(在grafana中效果相同).另外，我在bigtable的tsdb表中看不到任何数据.

When querying the data (GET on ./api/query), I don't see the data (same effect in grafana). Also, I do not see any data added in the tsdb table in bigtable.

我的结论是:尽管tsd返回204，但没有数据写入Bigtable.

My conclusion: no data is written to Bigtable, although tsd is returning 204.

有趣的事实:指标已创建(我可以在Bigtable(cbt read tsdb-uid)中看到它，还有opentsdb-ui(和grafana)中的自动完成功能立即就可以选择该指标.)没有数据.

Interesting fact: the metric is created (I can see it in Bigtable (cbt read tsdb-uid) and also the autocomplete in the opentsdb-ui (and grafana) pick the metric up right away. But no data.

当我在本教程中使用Heapster-Example时，一切都正常.

When I use the Heapster-Example as in the tutorial, it all works.

有趣的部分(对我来说):

And the interesting part (to me):

注意:发生了几次，有很大的延迟，或者在停止/重启kubernetes集群之后，数据才出现.突然.到目前为止，我无法复制.

NOTE: It happened a few times, with massive delay or after stoping/restarting the kubernetes cluster, that the data appeared. Suddenly. I could not reproduce as of now.

我一定错过了一些非常简单的东西.

I must be missing something really simple.

注意:我没有在日志(堆栈驱动程序)和UI(opentsdb UI)中看到任何错误，既没有bigtable也没有Kubernetes，也没有我能想到的东西.

Note: I don't see any errors in the logs (stackdriver) and UI (opentsdb UI), neither bigtable, nor Kubernetes, nor anything I can think of.

注意:我使用的配置与本教程中的链接相同.

Note: the configs I am using are as linked in the tutorial.

我正在使用的看跌期权(请参阅204):

The put I am using (see the 204):

，如果我添加?details，则表示成功:

and if I add ?details, it indicates success:

推荐答案

我的猜测是，这与opentsdb刷新频率有关.关闭tsdb群集时，将自动进行刷新.我不确定100％，但是我认为tsd.storage.flush_interval配置可以管理该过程.

My guess is that this relates to the opentsdb flush frequency. When a tsdb cluster is shutdown, there's an automatic flush. I'm not 100% sure, but I think that the tsd.storage.flush_interval configuration manages that process.

您可以通过google-cloud-bigtable-discuss组与维护图书馆的团队联系，您可以从

You can reach the team that maintains the libraries via the google-cloud-bigtable-discuss group, which you can get to from the Cloud Bigtable support page for more nuanced discussions.

作为一个仅供参考，我们(Google)正在积极更新 https://cloud .google.com/solutions/opentsdb-cloud-platform 到OpenTSDB和AsyncBigtable的最新版本，可以提高大容量的性能.

As an FYI, we (Google) are actively updating the https://cloud.google.com/solutions/opentsdb-cloud-platform to the latest versions of OpenTSDB and AsyncBigtable which should improve performance at high volumes.

这篇关于使用HTTP POST将OpenTSDB写入Bigtable不起作用(使用Kubernetes(的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！