RsRust Community
Community homeFeatured crates Release news Learning path

Rust Community · Six rules for error handling in Rust that I settled on

166
Rsr/rust·posted by rase·just nowTooling

Six rules for error handling in Rust that I settled on

Some background first. Our setup is Rust plus three downstream services, seven figures of daily requests, peaking around nine in the evening.

We also fixed monitoring along the way: replaced average-based alerts with percentiles and split them per endpoint. False alerts dropped by about seventy percent and the on-call rotation visibly cheered up.

What genuinely surprised me was the tail. The average looked great while P99 jumped by an order of magnitude past some threshold. The cause was not Rust itself but our upstream connection reuse — the load test traffic was too clean and hid the long-tail requests.

Worth noting: the official docs do cover this, just in a very inconspicuous spot. I only found it reading the source comments, where the author explains the reasoning — roughly "so that it degrades into predictable behaviour in extreme cases".

One last trap: in container environments remember to adjust the memory-related parameters in step. Otherwise the host limit and the process expectation disagree, and the symptom is intermittent, unreproducible failure.

67 comments

67 comments

M
Ttang_hao·12 minutes ago

Worth learning from this debugging approach. We went straight at the logs and took a much longer route.

445
Cchen_dev·just now

Sharing our numbers, 8 cores 16GB, same scenario:

| Concurrency | P50 | P99 |
|---|---|---|
| 200 | 12ms | 88ms |
| 500 | 31ms | 340ms |

P99 clearly collapses at 500 concurrency, which lines up with your knee point.

464
Sswoole_leeMod·1 hour ago

This matches what we see in production. We only hit it past 3k QPS; the earlier load tests showed nothing — the test traffic was too clean, with no long-tail requests.

387
Kkernel_panic·2 days ago

Can you give a minimal reproduction? I ran it locally for ten minutes and could not reproduce on macOS with the latest version.

89
Zzhu_zong·12 minutes ago

Has anyone run a controlled experiment? I did, reducing it to a single variable, and the difference was 4% — within noise. So I suspect the main cause is something else.

237
Oops_wang·12 minutes agoLevel 6

We have run this in production for two years without hitting it. That said, we never reached this scale, so our experience is not really evidence here.

83
Kkernel_panic·2 hours agoedited

One counter-example: below Rust 7.4 the semantics of that code are different, so do not copy it verbatim. We got burned in staging and rolled back once.

115
Kkernel_panic·2 days agoLevel 6

Thanks for sharing real numbers — far more useful than the articles that only cover concepts.

1
Cchen_dev·1 hour ago

Saved. I am reworking this area this week — this saves a lot of wrong turns.

81
Ddev_zhou·2 days ago

One counter-example: below Rust 7.4 the semantics of that code are different, so do not copy it verbatim. We got burned in staging and rolled back once.

166
Wwinter·2 days ago

A question: what changes in a container with a 512Mi memory limit? That is how we run it in production.

122
Ttang_hao·2 days ago

A question: what changes in a container with a 512Mi memory limit? That is how we run it in production.

51
Kkernel_panic·2 days ago

Saved. I am reworking this area this week — this saves a lot of wrong turns.

50
Ddev_zhou·2 days ago

There is actually a simpler fix that needs no architecture change: move this check up to the gateway and the problem disappears. The cost is one extra lookup at the gateway.

2
Sslow_query·2 days ago

We have run this in production for two years without hitting it. That said, we never reached this scale, so our experience is not really evidence here.

106
Zzhou_yi·28 minutes ago

Thanks for sharing real numbers — far more useful than the articles that only cover concepts.

173
Ttang_haoMod·2 days ago

Sharing our numbers, 8 cores 16GB, same scenario:

| Concurrency | P50 | P99 |
|---|---|---|
| 200 | 12ms | 88ms |
| 500 | 31ms | 340ms |

P99 clearly collapses at 500 concurrency, which lines up with your knee point.

488
Nnikic·2 days ago

One counter-example: below Rust 7.4 the semantics of that code are different, so do not copy it verbatim. We got burned in staging and rolled back once.

355
Rrase·2 days agoLevel 6

There is actually a simpler fix that needs no architecture change: move this check up to the gateway and the problem disappears. The cost is one extra lookup at the gateway.

454
Cchen_devMod·2 days ago

Can you give a minimal reproduction? I ran it locally for ten minutes and could not reproduce on macOS with the latest version.

40
Sswoole_lee·5 hours ago

Has anyone run a controlled experiment? I did, reducing it to a single variable, and the difference was 4% — within noise. So I suspect the main cause is something else.

171
Rran_boOP·2 days ago

This is not a Rust problem, it is a usage problem. The docs say this API is not thread-safe and you must lock around it yourself.

415
Ttang_hao·2 hours ago

I see point 3 differently. The trade-off depends on your read/write ratio: read-heavy with little writing means caching actually widens the inconsistency window.

193
Zzhou_yi·5 hours ago

I see point 3 differently. The trade-off depends on your read/write ratio: read-heavy with little writing means caching actually widens the inconsistency window.

11
Oops_wang·3 minutes ago

Can you give a minimal reproduction? I ran it locally for ten minutes and could not reproduce on macOS with the latest version.

431
Oops_wangMod·2 days ago

Worth learning from this debugging approach. We went straight at the logs and took a much longer route.

7
Rrase·2 days ago

Saved. I am reworking this area this week — this saves a lot of wrong turns.

414
Wwinter·just now

I see point 3 differently. The trade-off depends on your read/write ratio: read-heavy with little writing means caching actually widens the inconsistency window.

1
Mmike_xu·2 days ago

We have run this in production for two years without hitting it. That said, we never reached this scale, so our experience is not really evidence here.

139
Kkernel_panic·2 days ago

I see point 3 differently. The trade-off depends on your read/write ratio: read-heavy with little writing means caching actually widens the inconsistency window.

392
Mmike_xu·2 days ago

There is actually a simpler fix that needs no architecture change: move this check up to the gateway and the problem disappears. The cost is one extra lookup at the gateway.

228
Wwinter·2 days agoLevel 6

This is not a Rust problem, it is a usage problem. The docs say this API is not thread-safe and you must lock around it yourself.

209
Kkernel_panicOP·2 days agoedited

I just read the Rust source — the author actually explains the reasoning in a comment, roughly "so that it degrades into predictable behaviour in extreme cases".

1
Mmike_xu·5 hours ago

I just read the Rust source — the author actually explains the reasoning in a comment, roughly "so that it degrades into predictable behaviour in extreme cases".

5
Hhuang_ke·just now

I just read the Rust source — the author actually explains the reasoning in a comment, roughly "so that it degrades into predictable behaviour in extreme cases".

343
Rrase·12 minutes ago

Agreeing with the above. One addition: with this option enabled the GC count in your metrics doubles, so adjust the alert threshold at the same time or it will keep firing.

261
Kkernel_panic·28 minutes agoedited

Thanks for sharing real numbers — far more useful than the articles that only cover concepts.

230
Hhuang_keMod·2 days ago

A question: what changes in a container with a 512Mi memory limit? That is how we run it in production.

458
Cchen_dev·2 days ago

Saved. I am reworking this area this week — this saves a lot of wrong turns.

455
Wwinter·2 days ago

Agreeing with the above. One addition: with this option enabled the GC count in your metrics doubles, so adjust the alert threshold at the same time or it will keep firing.

175
Lli_ming·2 days ago

Has anyone run a controlled experiment? I did, reducing it to a single variable, and the difference was 4% — within noise. So I suspect the main cause is something else.

190
Kkite·5 hours ago

There is actually a simpler fix that needs no architecture change: move this check up to the gateway and the problem disappears. The cost is one extra lookup at the gateway.

99
Sswoole_lee·2 days ago

This matches what we see in production. We only hit it past 3k QPS; the earlier load tests showed nothing — the test traffic was too clean, with no long-tail requests.

249
Zzhou_yi·2 days agoedited

Has anyone run a controlled experiment? I did, reducing it to a single variable, and the difference was 4% — within noise. So I suspect the main cause is something else.

465
Aalice_dev·2 days ago

Worth learning from this debugging approach. We went straight at the logs and took a much longer route.

199
Nnikic·just now

Saved. I am reworking this area this week — this saves a lot of wrong turns.

185
Kkernel_panic·2 days agoedited

This is not a Rust problem, it is a usage problem. The docs say this API is not thread-safe and you must lock around it yourself.

257
WwinterOP·2 days ago

Agreeing with the above. One addition: with this option enabled the GC count in your metrics doubles, so adjust the alert threshold at the same time or it will keep firing.

15
KkiteOP·2 days ago

This is not a Rust problem, it is a usage problem. The docs say this API is not thread-safe and you must lock around it yourself.

305
Mmike_xu·2 days ago

A question: what changes in a container with a 512Mi memory limit? That is how we run it in production.

19
Zzhu_zong·yesterday

I just read the Rust source — the author actually explains the reasoning in a comment, roughly "so that it degrades into predictable behaviour in extreme cases".

2
Kkernel_panic·2 hours ago

This is not a Rust problem, it is a usage problem. The docs say this API is not thread-safe and you must lock around it yourself.

1
Ddev_zhouOP·2 days ago

Thanks for sharing real numbers — far more useful than the articles that only cover concepts.

103
Ddev_zhou·2 days ago

Sharing our numbers, 8 cores 16GB, same scenario:

| Concurrency | P50 | P99 |
|---|---|---|
| 200 | 12ms | 88ms |
| 500 | 31ms | 340ms |

P99 clearly collapses at 500 concurrency, which lines up with your knee point.

123
Kkite·28 minutes agoeditedLevel 6

This matches what we see in production. We only hit it past 3k QPS; the earlier load tests showed nothing — the test traffic was too clean, with no long-tail requests.

366
Kkernel_panic·12 minutes ago

I see point 3 differently. The trade-off depends on your read/write ratio: read-heavy with little writing means caching actually widens the inconsistency window.

147
Bbob_chen·2 days ago

We have run this in production for two years without hitting it. That said, we never reached this scale, so our experience is not really evidence here.

131
Rran_bo·28 minutes ago

A question: what changes in a container with a 512Mi memory limit? That is how we run it in production.

112
Sswoole_lee·2 days ago

Sharing our numbers, 8 cores 16GB, same scenario:

| Concurrency | P50 | P99 |
|---|---|---|
| 200 | 12ms | 88ms |
| 500 | 31ms | 340ms |

P99 clearly collapses at 500 concurrency, which lines up with your knee point.

87
Zzhou_yiOP·2 hours ago

This matches what we see in production. We only hit it past 3k QPS; the earlier load tests showed nothing — the test traffic was too clean, with no long-tail requests.

1
Sswoole_lee·2 days ago

We have run this in production for two years without hitting it. That said, we never reached this scale, so our experience is not really evidence here.

30
Bbob_chen·1 hour agoedited

Can you give a minimal reproduction? I ran it locally for ten minutes and could not reproduce on macOS with the latest version.

7
Sslow_query·just now

One counter-example: below Rust 7.4 the semantics of that code are different, so do not copy it verbatim. We got burned in staging and rolled back once.

3
Bbob_chen·2 days ago

This matches what we see in production. We only hit it past 3k QPS; the earlier load tests showed nothing — the test traffic was too clean, with no long-tail requests.

3
Zzhou_yi·just now

Agreeing with the above. One addition: with this option enabled the GC count in your metrics doubles, so adjust the alert threshold at the same time or it will keep firing.

318
Kkite·2 days ago

I just read the Rust source — the author actually explains the reasoning in a comment, roughly "so that it degrades into predictable behaviour in extreme cases".

260
Rran_bo·5 hours agoedited

Worth learning from this debugging approach. We went straight at the logs and took a much longer route.

1

This is the post detail page /en/c/rust/post/p8. Posts and comments are generated deterministically from a seeded PRNG, so the same post always renders the same content and the link can be shared, reloaded and indexed. In production this page reads MySQL for the post, Redis for hot-post caching, and fetches the whole comment tree in a single query on the path column.

See the database schema →