Slow Application Staging After Platform Upgrades

Products

• VMware Tanzu Application Service

Issue/Introduction

Following a platform upgrade or any event where Diego Cells are recreated, users may experience significantly longer wait times during the initial application staging process.

In Tanzu Elastic Application Runtime, buildpacks are stored centrally in the Blobstore. To optimize performance, Diego Cells maintain a local cache of these buildpacks. When cells are updated or recreated, this local cache is cleared. The first "push" or staging request handled by a cell after an update requires a full download of the necessary buildpack from the Blobstore, leading to a temporary spike in staging duration.

Resolution

There is currently no native "auto-sync" or "pre-populate" mechanism to push buildpacks to all Diego Cells simultaneously. The cache is populated "on-demand" only when a cell is assigned a staging task.

The most common root cause for excessive staging delays is low throughput between the Diego Cells and the Blobstore (whether internal or external). If the download speed is significantly lower than your enterprise network's capability, it indicates a bottleneck. We recommend reviewing the network path to identify:

• Bandwidth throttling or rate-limiting.

• High latency between the Diego Cell subnet and the Blobstore endpoint.

• Misconfigured MTU settings or firewall inspection delays.