International Business Machines Corporation

Biao Chai

Yue Wang

Xin Peng Liu

Liang Wang

Wei Wu

Peng Hui Jiang

Zheng Li

Shuang Men

Xinpeng Liu

A method includes, in response to receiving an incoming service request and establishing a call chain of pods of a service mesh network, setting a retry locker parameter to a locked state for each pod in the call chain. A locked retry locker parameter prevents the pod from initiating retries of a service request. The method includes, in response to determining that a pod in the call chain is unavailable, setting the retry locker parameter to an unlocked state for a previous pod just prior to the pod that is unavailable. The unlocked state allows a retry to the pod that is unavailable. In response to the previous pod reaching a retry limit, the method includes setting the retry locker parameter to unlocked for each pod in the call chain and sending a service termination message to a service requester.

Approach to gracefully resolve retry storms

Methods for managing a connection pool for cloud-based applications accessing databases are provided. Aspects include establishing a plurality of connections to one or more databases, receiving connection requests from a plurality of cloud-based applications to access the one or more databases, and allocating the connection requests to among the plurality of connections, where the allocation is determined based at least in part on one or more connection pool parameters. Aspects also include monitoring a usage of the plurality of connections by the plurality of applications and updating the one or more connection pool parameters based on an analysis of the usage of the plurality of connections by the plurality of applications.

Managing a connection pool for cloud-based applications accessing databases

Managing application migration is provided. An API server on a controller node is invoked to update a configuration map of a reverse proxy on a worker node for the reverse proxy to route user service requests corresponding to unmigrated applications of a set of applications from a first computing platform to a second computing platform to maintain liveness of the unmigrated applications during migration. The API server is invoked to build an image for an application of the set of applications based on source code of the application obtained from the second computing platform. The API server is invoked to generate a pod on the worker node to perform a workload of the application using the image. The API server is invoked to update a service on the worker node to select the pod on the worker node performing the workload of the application.

Continuous liveness and integrity of applications during migration

Selective privileged container augmentation is provided. A target group of edge devices is selected from a plurality of edge devices to run a plurality of child tasks comprising a pending task by mapping edge device tag attributes of the plurality of edge devices to child task tag attributes of the plurality of child tasks. A privileged container corresponding to the pending task is installed in each edge device of the target group to monitor execution of a child task by a given edge device of the target group. A privileged container installation tag that corresponds to the privileged container is added to an edge device tag attribute of each edge device of the target group having the privileged container installed. A child task of the plurality of child tasks comprising the pending task is sent to a selected edge device in the target group to run the child task.

Selective privileged container augmentation

Dynamically verifying ingress configuration changes is provided. A temporary ingress controller configuration is generated for an ingress configuration change set contained in an ingress configuration change set dispatcher queue of an ingress controller pod. The temporary ingress controller configuration corresponding to the ingress configuration change set is loaded into a temporary ingress controller located in a temporary ingress controller pod of the computer. A health check is performed on the temporary ingress controller pod running the temporary ingress controller with the temporary ingress controller configuration corresponding to the ingress configuration change set. The temporary ingress controller configuration is used as a configuration for an ingress controller located in the ingress controller pod of the computer based on the health check indicating that the temporary ingress controller pod did not crash while running the temporary ingress controller with the temporary ingress controller configuration corresponding to the ingress configuration change set.

Growing community of inventors

Beijing, China

Biao Chai