phpPHP Community
Community homeNews Release notes PHP wiki Web directory Multi-levelFeatured packages Best practices Q&A Jobs 38

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

148
phpr/php·posted by mike_xu·2 days agoPostmortemLocked

Six rules for error handling in PHP that I settled on

Short version: PHP 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.

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".

-- The query that broke: a full scan over 20M rows.
-- A composite index took P99 from 1.8s down to 42ms.
SELECT id, title, created_at
  FROM posts
 WHERE community_id = ?
   AND status = 1
 ORDER BY score DESC
 LIMIT 20;

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.

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.

65 comments

65 comments

M
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.

472
Cchen_dev·2 days 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.

69
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.

472
Cchen_dev·2 days ago

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

333
Sswoole_lee·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.

47
Hhuang_ke·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.

9
Rran_bo·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.

452
Llinlin·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.

299
Llinlin·2 days agoedited

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.

159
Aalice_devOP·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.

239
NnikicMod·28 minutes ago

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

273
Sswoole_lee·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.

256
Zzhou_yi·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.

118
Aalice_devOP·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.

184
Sslow_queryOP·12 minutes ago

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

173
Hhuang_ke·just now

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

230
Oops_wang·2 days ago

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

138
Cchen_dev·2 days ago

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

192
Sswoole_leeOP·just now

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

127
Mmike_xu·2 hours ago

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

182
Oops_wang·2 days ago

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

385
Zzhu_zong·1 hour ago

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

176
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.

472
Ddev_zhouOP·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.

9
Oops_wang·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.

31
Mmike_xu·yesterday

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.

79
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.

28
Llinlin·2 days ago

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

389
Llinlin·2 days ago

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

4
Mmike_xu·1 hour ago

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

32
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.

30
Kkite·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.

159
Bbob_chen·2 days ago

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

135
Rrase·just now

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

110
Llinlin·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.

346
Kkite·5 hours agoedited

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

284
Ttang_hao·yesterdayedited

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

4
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.

143
Wwinter·just now

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

208
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.

32
NnikicOP·2 days ago

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

34
Zzhou_yi·2 days ago

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

79
Sslow_queryOP·12 minutes ago

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

9
Ddev_zhou·3 minutes ago

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

291
Hhuang_ke·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.

298
Rrase·just nowedited

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.

167
NnikicOP·2 days ago

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

215
Ttang_hao·2 days ago

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

89
Ttang_hao·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.

57
Kkernel_panic·2 days ago

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

49
Nnikic·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.

39
Sswoole_lee·28 minutes ago

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

265
Mmike_xu·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.

36
Lli_ming·2 days ago

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

34
Hhuang_ke·2 hours 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.

22
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.

11
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.

32
Rrase·yesterday

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

10
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.

231
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.

173
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.

7
Rrase·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.

6
Ttang_hao·12 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.

1
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.

71
Wwinter·2 days ago

This is not a PHP 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/php/post/p5. 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 →