MagicMismatch warning

[debajyoti-truefoundry]

query-server
2025-06-17T12:33:55.042032Z  WARN                        main ThreadId(01) foyer_storage::large::generic: [lodc load]: deserialize read buffer raise error, remove this entry and skip hash=4942677581900972571 addr=EntryAddress { region: 0, offset: 201269248, len: 2610575, sequence: 4737 } e=MagicMismatch { expected: 2533566208, get: 1102495744 }
query-server
2025-06-17T12:33:55.131449Z  WARN                        main ThreadId(01) foyer_storage::large::generic: [lodc load]: deserialize read buffer raise error, remove this entry and skip hash=895246485322301758 addr=EntryAddress { region: 0, offset: 948928512, len: 1299828, sequence: 5253 } e=MagicMismatch { expected: 2533566208, get: 134226432 }
query-server
2025-06-17T12:33:55.354251Z  WARN                        main ThreadId(01) foyer_storage::large::generic: [lodc load]: deserialize read buffer raise error, remove this entry and skip hash=5128879209827795075 addr=EntryAddress { region: 0, offset: 960626688, len: 5669866, sequence: 5260 } e=MagicMismatch { expected: 2533566208, get: 808684544 }
query-server
2025-06-17T12:33:55.400697Z  WARN                        main ThreadId(01) foyer_storage::large::generic: [lodc load]: deserialize read buffer raise error, remove this entry and skip hash=4281808468104670649 addr=EntryAddress { region: 0, offset: 213487616, len: 1429856, sequence: 4745 } e=MagicMismatch { expected: 2533566208, get: 1225261056 }
query-server
2025-06-17T12:33:55.601630Z  WARN                        main ThreadId(01) foyer_storage::large::generic: [lodc load]: deserialize read buffer raise error, remove this entry and skip hash=13798441379288062264 addr=EntryAddress { region: 0, offset: 225243136, len: 6355730, sequence: 4754 } e=MagicMismatch { expected: 2533566208, get: 51437568 }
query-server
2025-06-17T12:33:55.652643Z  WARN                        main ThreadId(01) foyer_storage::large::generic: [lodc load]: deserialize read buffer raise error, remove this entry and skip hash=15997015890639037484 addr=EntryAddress { region: 0, offset: 971218944, len: 4939365, sequence: 5266 } e=MagicMismatch { expected: 2533566208, get: 1972073728 }
query-server
2025-06-17T12:33:55.868956Z  WARN                        main ThreadId(01) foyer_storage::large::generic: [lodc load]: deserialize read buffer raise error, remove this entry and skip hash=15082129487603746370 addr=EntryAddress { region: 0, offset: 231600128, len: 350702, sequence: 4755 } e=MagicMismatch { expected: 2533566208, get: 2169767680 }
query-server
2025-06-17T12:33:55.925307Z  WARN                        main ThreadId(01) foyer_storage::large::generic: [lodc load]: deserialize read buffer raise error, remove this entry and skip hash=4874503669074447421 addr=EntryAddress { region: 0, offset: 981078016, len: 2633408, sequence: 5272 } e=MagicMismatch { expected: 2533566208, get: 1630759936 }
query-server
2025-06-17T12:33:56.044095Z  WARN                        main ThreadId(01) foyer_storage::large::generic: [lodc load]: deserialize read buffer raise error, remove this entry and skip hash=12571071698983051371 addr=EntryAddress { region: 0, offset: 231952384, len: 1398591, sequence: 4760 } e=MagicMismatch { expected: 2533566208, get: 387592192 }

Hello. I am using 0.17.3.

1. What do the above warnings mean?
2. Where can I read up more about lodc?

[MrCroxx]

Hi. Thanks for asking.

This warning can be triggered by various reasons, but none of them is a big deal:

1. foyer pre-allocate file space on init to reduce runtime overhead. On recovery, if foyer tries to detect if a region has data, this warning can be triggered if the region is not written and holds random data.
2. Last time foyer didn't close gracefully. The last batch of data of each flusher was partially written.
3. The last batch written by foyer was not synced (e.g. fsync(2), fdatasync(2)), and some data corrupted.

As a cache system, it is fine to lose some data if the ratio is low enough (at most the last batch). So don't worry about it.

Btw, do you feel the warning confusing? If so, I'll change it to debug level logging.

[debajyoti-truefoundry]

As a cache system, it is fine to lose some data if the ratio is low enough (at most the last batch). So don't worry about it.

Is this ratio, or the count of successful and unsuccessful operations exposed via Prometheus metrics?
I am already using,

        let shared_cache = HybridCacheBuilder::new()
            .with_metrics_registry(Box::new(PrometheusMetricsRegistry::new(
                prometheus_registry.clone(),
            )))

Btw, do you feel the warning confusing? If so, I'll change it to debug level logging.

If the ratio is supposed to be low, then the warning is fine.

Thank you very much for your prompt response.

[MrCroxx]

Is this ratio, or the count of successful and unsuccessful operations exposed via Prometheus metrics?

Yes. But no data for the recovery phase.

[debajyoti-truefoundry]

Is this ratio, or the count of successful and unsuccessful operations exposed via Prometheus metrics?

Yes. But no data for the recovery phase.

Thanks. Could you please point me to the Prometheus metric names for this?

[MrCroxx]

• foyer_storage_op_total with op=error
• foyer_storage_inner_op_total with op=buffer_overflow|channel_overflow