Skip to content

rofl-scheduler: Investigate memory usage during OCI image pulls #2384

@ptrus

Description

@ptrus

Testnet rofl-scheduler sometimes reached the 512MB limit while concurrently pulling 5+ OCI image layers. (current workaround: #2383)

We should also limit concurrency for this and not pull layers for all apps at once (e.g., on startup).

Some relevant logs:

{"instance_id":"00000000000003bd","level":"info","module":"runtime/scheduler/manager","msg":"pulling OCI image layers","ref":"rofl.sh/dd86bfee-7f3e-4f38-8ee0-14c33507c70c:1761900571@sha256:58fea041ff43efff4664181acb08a75da42b1f76813d1f528faf34d45a3652a3","temporary_name":"instance-00000000000003bd","ts":"2025-11-06T09:15:01.680937637Z"}
{"instance_id":"00000000000003ab","layer_name":"firmware.fd","level":"info","module":"runtime/scheduler/manager","msg":"pulling OCI layer","ref":"rofl.sh/d7d55e8f-f2a2-4917-8867-42d23727d54c:1761813307@sha256:2515ba4849b829a7dc73cdced60ff1e88b6bfed328c4f09a1497faf41da910d4","ts":"2025-11-06T09:15:01.692669744Z"}
{"instance_id":"0000000000000383","layer_name":"firmware.fd","level":"info","module":"runtime/scheduler/manager","msg":"pulling OCI layer","ref":"rofl.sh/f3e9b5d4-89e4-41d8-9650-7e29908ef7c7:1761138620@sha256:31d8330edfe29dd535f0a467156797f113baaa5d826a99457baf0ec2e754da52","ts":"2025-11-06T09:15:01.704687133Z"}
{"instance_id":"00000000000003bd","layer_name":"firmware.fd","level":"info","module":"runtime/scheduler/manager","msg":"pulling OCI layer","ref":"rofl.sh/dd86bfee-7f3e-4f38-8ee0-14c33507c70c:1761900571@sha256:58fea041ff43efff4664181acb08a75da42b1f76813d1f528faf34d45a3652a3","ts":"2025-11-06T09:15:01.716654472Z"}
{"instance_id":"000000000000035d","layer_name":"kernel.bin","level":"info","module":"runtime/scheduler/manager","msg":"pulling OCI layer","ref":"rofl.sh/3bd165af-2981-4601-96b6-890397d37556:1760461048@sha256:5edadc67bc9c59de3bd7aa15ce483a73dda3c5826e7d7e259f627c3f46f98000","ts":"2025-11-06T09:15:02.472870258Z"}
{"instance_id":"00000000000003a9","layer_name":"kernel.bin","level":"info","module":"runtime/scheduler/manager","msg":"pulling OCI layer","ref":"rofl.sh/54699ba7-3c32-47de-841a-6781544fb12d:1761647823@sha256:63caa8fc806605a776927939244bb5303ad0ac3ddad85f0b7b4942b85aba25b6","ts":"2025-11-06T09:15:02.566545569Z"}
{"instance_id":"00000000000003c8","layer_name":"kernel.bin","level":"info","module":"runtime/scheduler/manager","msg":"pulling OCI layer","ref":"rofl.sh/44109cf3-3da9-45a9-b9b4-4718be757e56:1762017912@sha256:74966595dd18bf958c1030b4cc83038f88b065c388b9bc933c9eec6af808ccd7","ts":"2025-11-06T09:15:03.236934474Z"}
{"instance_id":"00000000000003ab","layer_name":"kernel.bin","level":"info","module":"runtime/scheduler/manager","msg":"pulling OCI layer","ref":"rofl.sh/d7d55e8f-f2a2-4917-8867-42d23727d54c:1761813307@sha256:2515ba4849b829a7dc73cdced60ff1e88b6bfed328c4f09a1497faf41da910d4","ts":"2025-11-06T09:15:03.352059506Z"}
{"instance_id":"0000000000000383","layer_name":"kernel.bin","level":"info","module":"runtime/scheduler/manager","msg":"pulling OCI layer","ref":"rofl.sh/f3e9b5d4-89e4-41d8-9650-7e29908ef7c7:1761138620@sha256:31d8330edfe29dd535f0a467156797f113baaa5d826a99457baf0ec2e754da52","ts":"2025-11-06T09:15:03.384580871Z"}
{"instance_id":"00000000000003bd","layer_name":"kernel.bin","level":"info","module":"runtime/scheduler/manager","msg":"pulling OCI layer","ref":"rofl.sh/dd86bfee-7f3e-4f38-8ee0-14c33507c70c:1761900571@sha256:58fea041ff43efff4664181acb08a75da42b1f76813d1f528faf34d45a3652a3","ts":"2025-11-06T09:15:03.398276786Z"}
{"instance_id":"000000000000035d","layer_name":"stage2.img","level":"info","module":"runtime/scheduler/manager","msg":"pulling OCI layer","ref":"rofl.sh/3bd165af-2981-4601-96b6-890397d37556:1760461048@sha256:5edadc67bc9c59de3bd7aa15ce483a73dda3c5826e7d7e259f627c3f46f98000","ts":"2025-11-06T09:15:05.619190134Z"}
{"instance_id":"00000000000003a9","layer_name":"stage2.img","level":"info","module":"runtime/scheduler/manager","msg":"pulling OCI layer","ref":"rofl.sh/54699ba7-3c32-47de-841a-6781544fb12d:1761647823@sha256:63caa8fc806605a776927939244bb5303ad0ac3ddad85f0b7b4942b85aba25b6","ts":"2025-11-06T09:15:06.340557002Z"}
{"instance_id":"00000000000003c8","layer_name":"stage2.img","level":"info","module":"runtime/scheduler/manager","msg":"pulling OCI layer","ref":"rofl.sh/44109cf3-3da9-45a9-b9b4-4718be757e56:1762017912@sha256:74966595dd18bf958c1030b4cc83038f88b065c388b9bc933c9eec6af808ccd7","ts":"2025-11-06T09:15:07.117864804Z"}
{"instance_id":"00000000000003ab","layer_name":"stage2.img","level":"info","module":"runtime/scheduler/manager","msg":"pulling OCI layer","ref":"rofl.sh/d7d55e8f-f2a2-4917-8867-42d23727d54c:1761813307@sha256:2515ba4849b829a7dc73cdced60ff1e88b6bfed328c4f09a1497faf41da910d4","ts":"2025-11-06T09:15:07.203221145Z"}
{"instance_id":"00000000000003bd","layer_name":"stage2.img","level":"info","module":"runtime/scheduler/manager","msg":"pulling OCI layer","ref":"rofl.sh/dd86bfee-7f3e-4f38-8ee0-14c33507c70c:1761900571@sha256:58fea041ff43efff4664181acb08a75da42b1f76813d1f528faf34d45a3652a3","ts":"2025-11-06T09:15:07.230679577Z"}
{"instance_id":"0000000000000383","layer_name":"stage2.img","level":"info","module":"runtime/scheduler/manager","msg":"pulling OCI layer","ref":"rofl.sh/f3e9b5d4-89e4-41d8-9650-7e29908ef7c7:1761138620@sha256:31d8330edfe29dd535f0a467156797f113baaa5d826a99457baf0ec2e754da52","ts":"2025-11-06T09:15:07.24309199Z"}
{"err":"Cannot allocate memory (os error 12)","level":"warn","module":"runtime/protocol","msg":"Failed to write message","ts":"2025-11-06T09:15:25.256614859Z"}
{"err":"failed to fill whole buffer","level":"error","module":"runtime/protocol","msg":"Failed to handle message","ts":"2025-11-06T09:15:25.283382097Z"}
{"level":"info","module":"runtime/protocol","msg":"Protocol reader thread is terminating","ts":"2025-11-06T09:15:25.293153009Z"}

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions