It looks like the app pool gets recycled, but the Ignite node is not shut down gracefully, causing the abrupt disconnect.
You can use code like this to stop all Ignite nodes gracefully when the application exits: AppDomain.CurrentDomain.ProcessExit += Ignition.StopAll(true); On Sat, Jan 29, 2022 at 7:18 AM Charlin S <[email protected]> wrote: > Hi, > > We are using IIS. > > Regards, > Charlin > > > On Fri, 28 Jan 2022 at 16:54, Pavel Tupitsyn <[email protected]> wrote: > >> ConsistentId is important for persistence, I don't think you should set >> it at all. >> "application pool" - do you use IIS or something like that? >> >> On Fri, Jan 28, 2022 at 1:48 PM Charlin S <[email protected]> wrote: >> >>> Hi, >>> >>> 1. Is there a reason to set it manually? >>> --We had node disconnection multiple times and were unable to join >>> back so we decided to set a new consistentId. >>> .Net 4.6.2 and .Net 5 WebAPI's are having the same logic. >>> 2. Do you use Ignite persistence? >>> -- No, our cache type is in-memory. >>> >>> Regards, >>> Charlin >>> >>> On Fri, 28 Jan 2022 at 12:57, Pavel Tupitsyn <[email protected]> >>> wrote: >>> >>>> Hi Charlin, >>>> >>>> ConsistentId should not be random [1] >>>> >>>> 1. Is there a reason to set it manually? >>>> 2. Do you use Ignite persistence? >>>> >>>> [1] >>>> https://ignite.apache.org/docs/latest/persistence/native-persistence >>>> >>>> On Fri, Jan 28, 2022 at 6:53 AM Charlin S <[email protected]> >>>> wrote: >>>> >>>>> Hi Ignite team, >>>>> We are using Ignite 2.10.0 with .Net 5 WebAPI and we have a 16-nodes(2 >>>>> servers) Ignite cluster >>>>> The client nodes have the following property as below >>>>> IgniteConfiguration.AutoGenerateIgniteInstanceName = true; >>>>> IgniteConfiguration.ConsistentId = Guid.NewGuid().ToString().ToUpper(); >>>>> >>>>> Some of the client nodes are failing as "Node_Failed" during >>>>> application pool recycle every day, but it was different behaviour with >>>>> .Net 4.6.2 WebAPI as "Node Left Topology" >>>>> >>>>> I check the logs with the following findings: >>>>> Some nodes are marking as "Node Left Topology" and some nodes are >>>>> marking as "Node_Failed" >>>>> >>>>> We would like to figure out the root cause of the "Node_Failed" during >>>>> application pool recycle. >>>>> >>>>> >>>>> Regards, >>>>> Charlin >>>>> >>>>
