Startup of APIGateway stops with message

2021-03-18 22:24:13 CET [ISS.0035.0003I] Log rotation interval set to 1440 minutes.
2021-03-18 22:24:14 CET [ISS.0025.0057I] Reading system level information
2021-03-18 22:24:14 CET [ISS.0014.0002I] Initialization completed in 117 seconds.
2021-03-18 22:24:14 CET [ISS.0025.0016I] Config File Directory Saved
2021-03-18 22:24:24 CET [YAI.0001.9999E] [default][ulapit02.edv.dekra.de][APIGW:f01bfae9-d3e0-4ac7-b16b-4137a0de88f8:504:4407cc292b7b4e838b53d55332de6db0] Exception : Processing request for tenantInfo.gettenantId() failed. Error Message: org.springframework.context.support.GenericApplicationContext@2d0d9a37 has not been refreshed yet

Hi Gockel,
Can you please provide more information. Which version of API Gateway, are you unable to start it after installation? Are there any installation errors?

Regards.

Hi Prabavathy,
thank you for responding. This is APIGateway 10.5 with Fix 10 applied. This is new for us and the whole software is still under construction. Its about a cluster comprising 3 Linux nodes and each has Terracotta, Elasticsearch and IS/Apigateway installed on it. The problem occurs during shutting and startup of all processes on one arbitrarily selected node (this scenario anticipates a future maintenace window where the entire Linux Ldap needs a reboot). The problem does not always occur, e.g. 5 of 6 restarts function properly. But so far I fail to predict WHEN it recurr and WHY. When restarting I always take that the Terracotta component of the node beeing restarted has issued “Moved to State[ PASSIVE-STANDBY ]” BEFORE I continue with starting Elasticsearch and IS/ApiGateway. Also, a browser session with Terracotta management console is open to assure myself that an ACTIVE-MIRROR-MIRRO status quo is present. And to answer your 2nd question: no, installation/update problems evaded in conjunction with Installation- oder Update Manager. To resolve this situation an arduous proceeding is necessay: Shutdown all(!) nodes simulataneously, erase Terracottas …/work/terracotta/server-data/*, startup TC instances one by on, startup ES instances one by one and last the IS/Gateways one by one. This procedure includes a downtime and that’s make planning for production so difficult. Any idea ?
Thanks for your efforts - Hubert

Hi Gockel,
Below is the procedure to restart the nodes in a clustered deployment. Please let me know if you are seeing the issue after trying the below procedure.

Startup Sequence

  1. Start Terracotta
  2. Start Elasticsearch nodes
  3. Start API Gateway

Shutdown Sequence

  1. Stop API Gateway
  2. Stop Elasticsearch
  3. Stop Terracotta

Regards.

Hello Hi Prabavathy,
I always started and stopped the components in THAT VARY sequence. And as I wrote, this works about 5 times properly, and on the 6th time, without doing something differently, it fails.
It is unpredictable.
Regards Hubert