Re: Questions related to Autoscaler
Hi, zhanghao Thanks for your reply, it also looks good to me. Chen Zhanghao 于2023年8月11日周五 09:23写道： > Q1: if you use operator to submit a standalone mode job with reactive mode > enabled, KEDA should still work. > > Q2: For Flink versions, 1.17 is recommended, but 1.15 is also okay if you > backport the necessary changes listed in Autoscaler | Apache Flink > Kubernetes Operator > <https://nightlies.apache.org/flink/flink-kubernetes-operator-docs-release-1.6/docs/custom-resource/autoscaler/>. > For Kubernetes Operator, the latest stable version is 1.5 (1.6 is close but > not officially released yet), so stay on 1.5 is fine. > > Q3: The metrics monitored (as of v1.5) are: throughput, lag, busy time. > CPU and memory is not considered. And yes, backlog-processing.lag-threshold > is related to Kafka consumer lag, when job lag time is beyond this > threashold, autoscaler will prevent any downscaling behavior. > > > Best, > Zhanghao Chen > -- > *发件人:* Hou, Lijuan via user > *发送时间:* 2023年8月9日 3:04 > *收件人:* user@flink.apache.org > *主题:* Questions related to Autoscaler > > > Hi Flink team, > > > > This is Lijuan. I am working on our flink job to realize autoscaling. We > are currently using flink version of 1.16.1, and using flink operator > version of 1.5.0. I have some questions need to confirm with you. > > > > 1 - It seems for flink job using flink operator to realize autoscaling, > the only option to realize autoscaling is to enable the Autoscaler feature, > and KEDA won’t work, right? > > > > 2 - I noticed from the document that we need to upgrade to flink version > of 1.17 to use Autoscaler. But I also noticed that the updated version for > flink operator is 1.7 now. > > Shall we upgrade from 1.5.0 to 1.7 to enable Autoscaler? > > > > 3 – I have done a lot of search, and also read the Autoscaler Algorithm > page. But I am still not very sure about the list of metrics observed > automatically. > >- Will it include CPU load, memory, throughput and kafka consumer lag? >Could you please provide the whole list of monitored metrics? > > - Is this config related to kafka consumer lag? > kubernetes.operator.job.autoscaler.backlog-processing.lag-threshold Thanks > a lot for the help! Best, Lijuan > > > > > >
回复: Questions related to Autoscaler
Q1: if you use operator to submit a standalone mode job with reactive mode enabled, KEDA should still work. Q2: For Flink versions, 1.17 is recommended, but 1.15 is also okay if you backport the necessary changes listed in Autoscaler | Apache Flink Kubernetes Operator<https://nightlies.apache.org/flink/flink-kubernetes-operator-docs-release-1.6/docs/custom-resource/autoscaler/>. For Kubernetes Operator, the latest stable version is 1.5 (1.6 is close but not officially released yet), so stay on 1.5 is fine. Q3: The metrics monitored (as of v1.5) are: throughput, lag, busy time. CPU and memory is not considered. And yes, backlog-processing.lag-threshold is related to Kafka consumer lag, when job lag time is beyond this threashold, autoscaler will prevent any downscaling behavior. Best, Zhanghao Chen 发件人: Hou, Lijuan via user 发送时间: 2023年8月9日 3:04 收件人: user@flink.apache.org 主题: Questions related to Autoscaler Hi Flink team, This is Lijuan. I am working on our flink job to realize autoscaling. We are currently using flink version of 1.16.1, and using flink operator version of 1.5.0. I have some questions need to confirm with you. 1 - It seems for flink job using flink operator to realize autoscaling, the only option to realize autoscaling is to enable the Autoscaler feature, and KEDA won’t work, right? 2 - I noticed from the document that we need to upgrade to flink version of 1.17 to use Autoscaler. But I also noticed that the updated version for flink operator is 1.7 now. Shall we upgrade from 1.5.0 to 1.7 to enable Autoscaler? 3 �C I have done a lot of search, and also read the Autoscaler Algorithm page. But I am still not very sure about the list of metrics observed automatically. * Will it include CPU load, memory, throughput and kafka consumer lag? Could you please provide the whole list of monitored metrics? - Is this config related to kafka consumer lag? kubernetes.operator.job.autoscaler.backlog-processing.lag-threshold Thanks a lot for the help! Best, Lijuan
Questions related to Autoscaler
Hi Ron, Thanks for the reply! > 1 - It seems for flink job using flink operator to realize autoscaling, the > only option to realize autoscaling is to enable the Autoscaler feature, and > KEDA won’t work, right? What is KEDA mean? -> KEDA is a Kubernetes based Event Driven Autoscaler. I found some examples using Flink’s previous Reactive mode + KEDA to realize autoscaling. So if Autoscaler is enabled, is it still necessary to create KEDA resources? I think TaskManager instances are created and destroyed by the Flink JobManager now, and aren’t in a replication controller, so they can't be “scaled up” using traditional Kubernetes techniques like KEDA. Could you please help confirm? Thank you! -- > 2 - I noticed from the document that we need to upgrade to flink version of > 1.17 to use Autoscaler. But I also noticed that the updated version for flink > operator is 1.7 now. Shall we upgrade from 1.5.0 to 1.7 to enable Autoscaler? I have checked the flink-kubernetes-operator projection pom for release-1.5 branch, the dependency flink version is 1.16.1. So I recommend you update your flink-kubernetes-operator to 1.6. The latest stable release is 1.6. -> Thank you, so the dependency flink version of flink-kubernetes-operator version-1.6 is 1.17? Our current flink version is 1.16.1, so does it mean we need to: 1 – Update flink version to 1.17 2 – Update flink operator version to 1.6? Could you please help confirm? Thank you! -- Oh I have another question from the email, please have a look: 3 – Could you please provide a list of metrics observed by Autoscaler automatically? * Will it include CPU load, memory, throughput and kafka consumer lag? * Is there any configurations related to kafka consumer lag that we can setup to scale job by making Autoscaler monitor it? Like some threshold? Thanks a lot for the help!! Best, Lijuan
Re: Questions related to Autoscaler
Hi, Lijuan > 1 - It seems for flink job using flink operator to realize autoscaling, the only option to realize autoscaling is to enable the Autoscaler feature, and KEDA won’t work, right? What is KEDA mean? > 2 - I noticed from the document that we need to upgrade to flink version of 1.17 to use Autoscaler. But I also noticed that the updated version for flink operator is 1.7 now. Shall we upgrade from 1.5.0 to 1.7 to enable Autoscaler? I have checked the flink-kubernetes-operator projection pom for release-1.5 branch, the dependency flink version is 1.16.1. So I recommend you update your flink-kubernetes-operator to 1.6. The latest stable release is 1.6. Best, Ron Hou, Lijuan via user 于2023年8月9日周三 03:04写道： > Hi Flink team, > > > > This is Lijuan. I am working on our flink job to realize autoscaling. We > are currently using flink version of 1.16.1, and using flink operator > version of 1.5.0. I have some questions need to confirm with you. > > > > 1 - It seems for flink job using flink operator to realize autoscaling, > the only option to realize autoscaling is to enable the Autoscaler feature, > and KEDA won’t work, right? > > > > 2 - I noticed from the document that we need to upgrade to flink version > of 1.17 to use Autoscaler. But I also noticed that the updated version for > flink operator is 1.7 now. > > Shall we upgrade from 1.5.0 to 1.7 to enable Autoscaler? > > > > 3 – I have done a lot of search, and also read the Autoscaler Algorithm > page. But I am still not very sure about the list of metrics observed > automatically. > >- Will it include CPU load, memory, throughput and kafka consumer lag? >Could you please provide the whole list of monitored metrics? > > - Is this config related to kafka consumer lag? > kubernetes.operator.job.autoscaler.backlog-processing.lag-threshold Thanks > a lot for the help! Best, Lijuan > > > > > >
Questions related to Autoscaler
Hi Flink team, This is Lijuan. I am working on our flink job to realize autoscaling. We are currently using flink version of 1.16.1, and using flink operator version of 1.5.0. I have some questions need to confirm with you. 1 - It seems for flink job using flink operator to realize autoscaling, the only option to realize autoscaling is to enable the Autoscaler feature, and KEDA won’t work, right? 2 - I noticed from the document that we need to upgrade to flink version of 1.17 to use Autoscaler. But I also noticed that the updated version for flink operator is 1.7 now. Shall we upgrade from 1.5.0 to 1.7 to enable Autoscaler? 3 – I have done a lot of search, and also read the Autoscaler Algorithm page. But I am still not very sure about the list of metrics observed automatically. * Will it include CPU load, memory, throughput and kafka consumer lag? Could you please provide the whole list of monitored metrics? - Is this config related to kafka consumer lag? kubernetes.operator.job.autoscaler.backlog-processing.lag-threshold Thanks a lot for the help! Best, Lijuan
