0.18.36-v1

Vultr

Cluster placement

The vultr backend can now provision fleets with cluster placement.

type: fleet

nodes: 4
placement: cluster

resources:
  gpu: 8:MI300X

backends: [vultr]

Nodes in such a cluster will be interconnected and can be used to run distributed tasks.

Performance

The update optimizes the performance of dstack server, allowing a single server replica to handle up to 150 active runs, jobs, and instances. Capacity can be further increased by using PostgreSQL and running multiple server replicas.

Last, getting instance offers from backends when you run dstack apply has also been optimized and now takes less time.

What's changed

• Increase max active resources supported by server by @r4victor in dstackai/dstack#2189
• Implement bridge network mode for jobs by @un-def in dstackai/dstack#2191
• [Internal] Fix python-json-logger deprecation warning by @jvstme in dstackai/dstack#2201
• Fix local backend by @r4victor in dstackai/dstack#2203
• Implement offers cache by @r4victor in dstackai/dstack#2197
• Add /api/instances/list by @jvstme in dstackai/dstack#2199
• Allow getting by ID in /api/project/_/fleets/get by @jvstme in dstackai/dstack#2200
• Add termination reason and message to the runner API by @r4victor in dstackai/dstack#2204
• Add vpc cluster support in Vultr by @Bihan in dstackai/dstack#2196
• Fix instance_types not respected for pool instances by @r4victor in dstackai/dstack#2205
• Delete manually created empty fleets by @r4victor in dstackai/dstack#2206
• Return repo errors from runner by @r4victor in dstackai/dstack#2207
• Fix caching offers with GPU requirements by @jvstme in dstackai/dstack#2210
• Fix filtering idle instances by instance type by @jvstme in dstackai/dstack#2214
• Add more project URLs on PyPI by @jvstme in dstackai/dstack#2215

Full changelog: dstackai/dstack@0.18.35...0.18.36