Loki chunks size

Loki’s Azure “blobs” are really just the compressed chunks of log entries that the Ingester flushes to your <container>/chunks/<tenant-ID>/… path. The size of each blob is controlled by Loki itself, not by Azure:

• chunk_target_size (default \~1 MiB) tells Loki roughly how big to let a chunk grow before flushing.
• max_chunk_age (default 1 h) forces a flush once a chunk has been open for that long, even if it hasn’t hit chunk_target_size.
• chunk_idle_period (default 1 h) flushes any stream that hasn’t received new entries in that window. You tune these under your Ingester config (ingester.chunk_target_size, ingester.max_chunk_age, etc.) in loki.yaml or via Helm values (medium.com).

---

What Azure does under the covers

When Loki hands off a chunk file to Azure, it uses the Block Blob API. Azure will automatically break any upload into blocks (up to 50 000 blocks per blob) and stitch them together:

• The max block size and max blob size are governed by the Storage service version—e.g. service 2019-12-12+ supports up to 4000 MiB per block and \~190 TiB per blob (learn.microsoft.com).
• Loki’s client-side buffer controls (e.g. azure.upload_buffer_size, default 256 KiB) simply govern how much of the chunk is read/streamed per block upload; they do not change the logical size of your chunk blob (techdocs.akamai.com).

---

Bottom line

• To change blob sizes, adjust Loki’s chunk‐flush settings (chunk_target_size, max_chunk_age, etc.).
• You cannot alter Azure’s internal block-splitting behavior via Loki—it’s handled by the Azure SDK/service.
• The only other knobs are the SDK buffer sizes (upload_buffer_size, download_buffer_size), which affect upload/download performance but not the chunk‐blob’s overall size.