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

iOS · Open source · Discussion: is the ios-opensource ecosystem being replaced by another stack?

1.3K
IOr/ios-opensource·posted by zhu_zong·yesterdayAnnouncementPinned

Discussion: is the ios-opensource ecosystem being replaced by another stack?

It took me two weeks of on-and-off digging and plenty of wrong turns. Writing the process down as it happened so the next person spends less time.

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".

Self-host — full control44%
Managed service — less work28%
Hybrid: core self-hosted17%
Not decided yet11%

2128 votes total

521 comments

521 comments

· first 120 loaded
M
RraseOP·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.

468
Sswoole_leeOPMod·28 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.

443
Nnikic·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.

473
Sswoole_lee·2 days ago

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

470
Oops_wangOP·2 days ago

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

404
KkiteMod·2 days ago

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

454
Lli_ming·2 days ago

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

448
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.

240
Oops_wang·2 days ago

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

114
Aalice_dev·12 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.

400
Nnikic·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.

391
Cchen_dev·2 days ago

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

362
Lli_mingMod·just now

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.

356
Zzhou_yi·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.

34
Kkite·3 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.

5
Mmike_xuOP·2 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.

1
Zzhu_zong·2 hours agoedited

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

410
Rran_bo·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.

441
Llinlin·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.

265
Zzhu_zong·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.

82
Hhuang_ke·2 days ago

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

21
KkiteMod·2 days agoLevel 6

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

473
Sswoole_lee·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.

209
Hhuang_ke·2 days ago

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

20
Ttang_hao·3 minutes ago

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

39
Kkite·2 days ago

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

454
Aalice_dev·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.

335
Aalice_dev·28 minutes ago

One counter-example: below ios-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.

145
Aalice_dev·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.

365
Zzhou_yi·2 days ago

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

8
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.

6
Wwinter·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.

236
Rrase·2 days ago

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

218
Llinlin·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.

196
Kkernel_panicMod·2 days ago

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

129
Zzhu_zong·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.

182
Sswoole_leeOP·2 days ago

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

115
Bbob_chen·yesterday

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
Ttang_hao·2 days ago

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

127
Zzhu_zongOP·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.

59
Zzhu_zong·5 hours ago

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

108
Cchen_dev·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.

106
Nnikic·28 minutes agoedited

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

98
Nnikic·5 hours 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.

15
Aalice_dev·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.

412
Bbob_chen·1 hour 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.

155
Sslow_query·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.

506
Oops_wang·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.

24
Mmike_xuMod·2 days agoeditedLevel 6

One counter-example: below ios-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.

8
Ddev_zhou·yesterday

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.

401
Rrase·2 days ago

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

97
Oops_wang·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.

95
Sslow_query·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.

500
Lli_mingOP·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.

28
Cchen_dev·2 days ago

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

88
Nnikic·28 minutes ago

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

79
Ttang_hao·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.

72
Oops_wangOPMod·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.

5
Mmike_xu·2 days ago

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

61
Zzhou_yi·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.

49
Lli_ming·5 hours ago

One counter-example: below ios-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.

46
Zzhou_yi·12 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.

271
Cchen_dev·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.

17
Mmike_xu·just now

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

45
Rran_bo·3 minutes agoedited

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

305
Llinlin·5 hours ago

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

300
Kkite·2 days ago

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

53
Kkernel_panic·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.

15
Ddev_zhou·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.

162
Zzhou_yi·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.

6
Cchen_dev·1 hour 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.

2
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.

255
KkiteOP·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.

180
Oops_wang·2 days ago

One counter-example: below ios-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.

2
Oops_wangOP·2 days ago

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

426
Nnikic·2 days ago

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

22
Zzhu_zong·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.

319
Kkite·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.

1
Kkernel_panic·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.

54
Zzhu_zong·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.

171
Hhuang_ke·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.

28
Rrase·12 minutes agoedited

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.

50
Kkernel_panic·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.

281
Sslow_query·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.

265
Rrase·2 days ago

One counter-example: below ios-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.

8
Nnikic·2 days agoedited

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

30
Ttang_hao·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.

1
Oops_wang·1 hour ago

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

4
Rran_boOP·28 minutes agoedited

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.

98
Rrase·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.

333
Kkernel_panicMod·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.

1
Nnikic·2 days ago

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

51
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.

458
Mmike_xuOP·2 days agoLevel 6

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

92
Zzhou_yi·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.

208
Ddev_zhou·2 days ago

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

25
NnikicMod·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.

57
Aalice_dev·2 days ago

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

2
Llinlin·12 minutes ago

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

100
Oops_wang·2 days ago

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

455
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.

335
Aalice_dev·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.

90
Zzhou_yi·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.

1
Wwinter·2 days ago

One counter-example: below ios-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.

39
Wwinter·2 days ago

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

27
Kkernel_panic·2 days ago

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

243
Kkite·1 hour ago

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

1
Sswoole_lee·2 days ago

One counter-example: below ios-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.

517
Ddev_zhou·just now

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

16
Aalice_dev·2 days ago

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

12
Kkernel_panic·yesterday

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

8
KkiteMod·2 days ago

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

6
Rrase·yesterday

One counter-example: below ios-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.

5
Kkite·2 days ago

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

4
Rrase·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.

3
Bbob_chenMod·yesterday

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

2
Sslow_query·2 days ago

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

5
Sslow_query·12 minutes 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
Kkernel_panic·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.

2
Kkernel_panic·2 days ago

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

1

This is the post detail page /en/c/ios-opensource/post/p0. 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 →