OFOffline First · Open source
Community homeHot discussions Latest posts Best of Knowledge base Resources About

Offline First · Open source · Those easily-missed type details in offline-first-opensource

158
OFr/offline-first-opensource·posted by rase·1 hour agoExperience

Those easily-missed type details in offline-first-opensource

Short version: offline-first-opensource needs almost no tuning at small and medium scale — the point where it starts to hurt is much further out than most people assume. Full measurements below.

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 offline-first-opensource itself but our upstream connection reuse — the load test traffic was too clean and hid the long-tail requests.

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.

Jjvns.caExternal link · opens in a new tab
63 comments

63 comments

M
Sslow_query·just nowedited

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

401
Rrase·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.

14
Cchen_devOP·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.

308
Bbob_chen·2 days ago

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

494
Oops_wang·2 days ago

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

364
Sslow_query·2 days ago

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

44
Llinlin·yesterday

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

348
Mmike_xu·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.

58
Aalice_devOP·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.

280
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.

302
Mmike_xu·yesterday

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.

299
Hhuang_ke·2 days ago

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

288
Bbob_chen·28 minutes ago

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

147
Oops_wang·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.

1
Zzhou_yi·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.

266
Aalice_dev·28 minutes ago

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

497
Aalice_dev·5 hours ago

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

14
Zzhu_zong·2 days ago

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

12
Rran_bo·2 days ago

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

26
Ddev_zhou·3 minutes ago

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

153
Kkernel_panic·12 minutes agoedited

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

63
Mmike_xu·2 hours ago

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

67
Zzhou_yiOP·28 minutes 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

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

212
Mmike_xu·5 hours agoedited

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

8
Bbob_chen·1 hour 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.

19
Sslow_query·2 days ago

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

6
Aalice_dev·2 days ago

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

435
Kkernel_panic·2 days agoLevel 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.

122
Sswoole_lee·2 days agoLevel 6

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

7
Hhuang_ke·2 days agoedited

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

24
Oops_wang·yesterdayedited

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

18
Llinlin·2 days agoedited

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

146
Sswoole_lee·yesterday

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

129
Llinlin·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.

8
Zzhou_yi·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.

206
Kkernel_panicOP·2 days agoedited

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.

96
Kkernel_panic·1 hour ago

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

122
Zzhou_yi·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.

114
Wwinter·28 minutes agoedited

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.

68
Zzhu_zongOP·just nowedited

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.

7
KkiteOP·3 minutes ago

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

1
Llinlin·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.

48
Ddev_zhou·5 hours 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.

47
Lli_ming·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.

39
Rran_bo·2 days ago

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

3
Kkite·12 minutes ago

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

63
Zzhou_yi·3 minutes 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.

10
Zzhu_zong·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.

326
Sslow_query·2 days ago

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

151
Sslow_queryOP·1 hour 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.

397
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.

106
Kkernel_panic·2 days ago

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

2
Bbob_chen·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.

10
Ttang_hao·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.

7
Llinlin·2 days ago

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

6
Rran_bo·3 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.

1
Zzhou_yi·just now

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.

198
Hhuang_keOP·12 minutes 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.

291
Oops_wang·12 minutes ago

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

1
Ttang_hao·1 hour agoedited

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.

1
Zzhu_zong·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.

1
Rran_bo·2 days ago

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

294

This is the post detail page /en/c/offline-first-opensource/post/p12. 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 →