TATanstack · Ops
Community homeHot discussions Latest posts Best of Knowledge base Resources About

Tanstack · Ops · The edge cases the tanstack-ops docs never spell out

2.1K
TAr/tanstack-ops·posted by chen_dev·2 hours agoOpen source

The edge cases the tanstack-ops docs never spell out

Most tanstack-ops articles stop at "how to use it" and never cover "when not to use it". This is an attempt at the second half.

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.

Performance44%
Maintainability28%
Ecosystem and community17%
Hiring difficulty11%

3604 votes total

1063 comments

1063 comments

· first 120 loaded
M
Rran_bo·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.

519
Sswoole_leeOP·2 days ago

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

452
Llinlin·2 days agoedited

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.

504
Ddev_zhou·2 days ago

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

30
Nnikic·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.

498
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.

373
Wwinter·3 minutes ago

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

314
Rrase·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
Llinlin·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.

172
Zzhu_zong·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.

82
Hhuang_keOP·2 days ago

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

275
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.

335
Nnikic·1 hour agoedited

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.

283
Aalice_dev·just now

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

352
Aalice_dev·28 minutes ago

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

498
Oops_wang·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.

468
Nnikic·2 days ago

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

99
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.

85
Cchen_devOP·2 days ago

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

212
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.

56
Oops_wang·2 days agoLevel 6

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

42
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.

44
Zzhu_zong·2 days agoedited

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

512
Kkernel_panicOP·2 days ago

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

449
Ttang_hao·1 hour ago

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

229
Sslow_query·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.

391
KkiteOP·2 days agoLevel 6

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

22
Llinlin·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.

9
Lli_ming·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.

294
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.

446
Ddev_zhou·2 days agoLevel 6

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

254
RraseOP·2 days 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.

1
Rrase·2 days ago

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

229
Rrase·yesterdayLevel 6

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

14
RraseOP·2 days 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.

34
Ttang_hao·2 days ago

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

154
Hhuang_ke·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.

408
Hhuang_ke·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.

71
Kkernel_panic·2 days agoLevel 6

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.

354
Zzhu_zong·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.

153
Llinlin·2 days ago

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

141
Zzhou_yi·2 days ago

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

6
Wwinter·2 days ago

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

247
Oops_wangMod·3 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.

239
Zzhu_zongMod·2 days ago

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

102
Rrase·2 days ago

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

225
Hhuang_ke·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.

386
Bbob_chen·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.

211
Rrase·3 minutes ago

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

207
Kkite·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.

165
Sswoole_leeOP·2 days ago

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

236
Mmike_xu·2 days ago

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

434
Kkite·2 days ago

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

6
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.

92
Zzhou_yi·2 days ago

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

7
Rran_bo·2 days ago

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

65
Oops_wang·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.

97
Wwinter·2 days ago

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

1
Rran_bo·2 days agoedited

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

199
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.

162
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.

160
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.

149
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.

146
LlinlinMod·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.

138
Wwinter·just now

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

122
Oops_wang·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.

137
Kkernel_panic·yesterday

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

19
Aalice_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.

39
Lli_ming·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
Ddev_zhou·2 days agoedited

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

1
Zzhu_zong·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.

352
Cchen_devOP·2 days ago

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

166
Aalice_devOP·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.

3
Sswoole_lee·3 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.

45
Wwinter·3 minutes agoedited

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

196
Zzhu_zong·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.

178
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.

282
Sslow_query·2 days agoLevel 6

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

1
Cchen_dev·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
Wwinter·2 days agoLevel 6

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

77
Cchen_dev·2 days agoeditedLevel 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.

35
Sswoole_lee·2 days agoLevel 6

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
Rrase·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.

41
Mmike_xu·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.

196
Nnikic·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
Bbob_chenOP·2 days ago

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

66
Nnikic·2 days ago

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

75
Oops_wang·just now

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.

2
Aalice_devOP·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.

58
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.

102
WwinterMod·12 minutes ago

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

90
Lli_ming·just now

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

78
Nnikic·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
Cchen_devOP·2 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.

11
Ddev_zhou·2 days ago

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

61
Bbob_chen·2 days ago

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

68
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.

64
Zzhou_yiOP·just now

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

1
Sslow_query·12 minutes ago

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

298
Zzhu_zong·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.

162
Rran_bo·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
Nnikic·2 days ago

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

46
Lli_ming·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.

42
Ddev_zhou·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.

1
Rrase·yesterday

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

40
Rran_bo·2 days ago

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

35
Llinlin·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.

34
Ddev_zhou·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.

306
Bbob_chen·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.

34
Zzhu_zong·2 days ago

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

33
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.

27
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.

21
Ddev_zhou·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
Ddev_zhou·2 days ago

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

17
Lli_ming·2 days ago

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

14
Bbob_chen·2 days ago

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

10
Rrase·2 days ago

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

332
Rran_bo·2 days ago

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

4
Llinlin·28 minutes ago

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

1
Ddev_zhou·5 hours ago

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

1

This is the post detail page /en/c/tanstack-ops/post/p2. 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 →