LA QUESTION
Why does the same outage keep recurring?
L’une influence-t-elle fortement l’autre ? Les trois quarts des votes exprimés décident.
fermé · retour à la phase Actions
Mode invité. Choisissez un pseudo pour rejoindre la salle et participer.
Les questions sont réparties entre les groupes : vous recevez celles attribuées à votre groupe — pas toutes les 180.
tranchées : 176 / 180 à vous de répondre : 0 pourquoi
Personne ne pourrait répondre à plusieurs centaines de paires. Chaque paire va à un groupe ; une paire sur laquelle le groupe se divise est ensuite renvoyée à d’autres groupes, et peut donc apparaître plus tard dans votre liste même si elle a commencé ailleurs. Le nombre de paires tranchées inclut le travail de tous les groupes : c’est pourquoi il monte plus vite que votre propre liste ne diminue.
Dernier décompte : tranchées : 176 · transmises à des panels inter-groupes : 1 · contestées : 0 · sous le quorum : 3 — les panels ont besoin d’au moins max(2, 75 % du panel) votes exprimés ; votre plus petit panel compte 15 membres.
Retries are unbounded, so a slow dependency becomes a flood → influence ? → Feature flags are never cleaned up, so nobody knows which paths are live
Rien n’a encore été dit.
Phase terminée — l’archive des paires et des votes figure ci-dessus.Vérifier la logique ↗
| Reliability has no budget line of its own | → | Feature flags are never cleaned up, so nobody knows which paths are live | ouvert | discuter |
| The same three services cause most of the pages | → | … | ouvert | discuter |
| The incident channel fills with people asking for status instead of giving it | → | The service map in the wiki is two years old | ouvert | discuter |
| The staging environment does not resemble production | → | Rollbacks take longer than the outage they are meant to end | ouvert transmis | discuter |
| Tests that fail intermittently are retried until green | → | Customers report outages before monitoring does | ✓ oui | discuter |
| Alerts fire so often that on-call mutes them | → | Customers report outages before monitoring does | ✓ oui | discuter |
| Alerts fire so often that on-call mutes them | → | Every team has its own logging format | ✓ oui | discuter |
| Every team has its own logging format | → | Customers report outages before monitoring does | ✓ oui | discuter |
| The on-call rota has the same two people on it most weeks | → | Alerts fire so often that on-call mutes them | ✓ oui | discuter |
| Feature work always outranks reliability work in planning | → | Post-mortem actions are written up and then never scheduled | ✓ oui | discuter |
| One engineer knows how the payment path actually works | → | The on-call rota has the same two people on it most weeks | ✓ oui | discuter |
| The on-call rota has the same two people on it most weeks | → | Post-mortem actions are written up and then never scheduled | ✓ oui | discuter |
| Post-mortems assign actions to people who were not in the room | → | Post-mortem actions are written up and then never scheduled | ✓ oui | discuter |
| The service map in the wiki is two years old | → | The staging environment does not resemble production | ✓ oui | discuter |
| Rollbacks take longer than the outage they are meant to end | → | Reliability has no budget line of its own | ✓ oui | discuter |
| Reliability has no budget line of its own | → | The staging environment does not resemble production | ✓ oui | discuter |
| Retries are unbounded, so a slow dependency becomes a flood | → | The same three services cause most of the pages | ✓ oui | discuter |
| Reliability has no budget line of its own | → | The incident channel fills with people asking for status instead of giving it | ✓ oui | discuter |
| The same three services cause most of the pages | → | The staging environment does not resemble production | ✓ oui | discuter |
| The service map in the wiki is two years old | → | Feature flags are never cleaned up, so nobody knows which paths are live | ✓ oui | discuter |
| Feature flags are never cleaned up, so nobody knows which paths are live | → | Rollbacks take longer than the outage they are meant to end | ✓ oui | discuter |
| Feature flags are never cleaned up, so nobody knows which paths are live | → | The same three services cause most of the pages | ✓ oui | discuter |
| The on-call rota has the same two people on it most weeks | → | Customers report outages before monitoring does | ✓ déduit | discuter |
| Retries are unbounded, so a slow dependency becomes a flood | → | The staging environment does not resemble production | ✓ déduit | discuter |
| Post-mortem actions are written up and then never scheduled | → | Every team has its own logging format | ✗ non | discuter |
| Every team has its own logging format | → | Error budgets exist on a slide and nowhere else | ✗ non | discuter |
| Feature work always outranks reliability work in planning | → | Customers report outages before monitoring does | ✗ non | discuter |
| Error budgets exist on a slide and nowhere else | → | Tests that fail intermittently are retried until green | ✗ non | discuter |
| Error budgets exist on a slide and nowhere else | → | One engineer knows how the payment path actually works | ✗ non | discuter |
| Error budgets exist on a slide and nowhere else | → | Post-mortems assign actions to people who were not in the room | ✗ non | discuter |
| Customers report outages before monitoring does | → | Post-mortems assign actions to people who were not in the room | ✗ non | discuter |
| Post-mortem actions are written up and then never scheduled | → | Alerts fire so often that on-call mutes them | ✗ non | discuter |
| Post-mortems assign actions to people who were not in the room | → | Alerts fire so often that on-call mutes them | ✗ non | discuter |
| Error budgets exist on a slide and nowhere else | → | Customers report outages before monitoring does | ✗ non | discuter |
| Customers report outages before monitoring does | → | The on-call rota has the same two people on it most weeks | ✗ non | discuter |
| Post-mortem actions are written up and then never scheduled | → | Post-mortems assign actions to people who were not in the room | ✗ non | discuter |
| Feature work always outranks reliability work in planning | → | Tests that fail intermittently are retried until green | ✗ non | discuter |
| Tests that fail intermittently are retried until green | → | The on-call rota has the same two people on it most weeks | ✗ non | discuter |
| Post-mortem actions are written up and then never scheduled | → | The on-call rota has the same two people on it most weeks | ✗ non | discuter |
| Feature work always outranks reliability work in planning | → | Error budgets exist on a slide and nowhere else | ✗ non | discuter |
| Tests that fail intermittently are retried until green | → | Every team has its own logging format | ✗ non | discuter |
| The on-call rota has the same two people on it most weeks | → | Tests that fail intermittently are retried until green | ✗ non | discuter |
| Post-mortems assign actions to people who were not in the room | → | Every team has its own logging format | ✗ non | discuter |
| Feature work always outranks reliability work in planning | → | Post-mortems assign actions to people who were not in the room | ✗ non | discuter |
| Tests that fail intermittently are retried until green | → | Alerts fire so often that on-call mutes them | ✗ non | discuter |
| Post-mortems assign actions to people who were not in the room | → | Error budgets exist on a slide and nowhere else | ✗ non | discuter |
| Alerts fire so often that on-call mutes them | → | One engineer knows how the payment path actually works | ✗ non | discuter |
| Customers report outages before monitoring does | → | One engineer knows how the payment path actually works | ✗ non | discuter |
| Alerts fire so often that on-call mutes them | → | Post-mortem actions are written up and then never scheduled | ✗ non | discuter |
| Post-mortems assign actions to people who were not in the room | → | The on-call rota has the same two people on it most weeks | ✗ non | discuter |
| One engineer knows how the payment path actually works | → | Alerts fire so often that on-call mutes them | ✗ non | discuter |
| Customers report outages before monitoring does | → | Alerts fire so often that on-call mutes them | ✗ non | discuter |
| Error budgets exist on a slide and nowhere else | → | The on-call rota has the same two people on it most weeks | ✗ non | discuter |
| Feature work always outranks reliability work in planning | → | Every team has its own logging format | ✗ non | discuter |
| Customers report outages before monitoring does | → | Post-mortem actions are written up and then never scheduled | ✗ non | discuter |
| Customers report outages before monitoring does | → | Every team has its own logging format | ✗ non | discuter |
| Customers report outages before monitoring does | → | Error budgets exist on a slide and nowhere else | ✗ non | discuter |
| Post-mortem actions are written up and then never scheduled | → | One engineer knows how the payment path actually works | ✗ non | discuter |
| Customers report outages before monitoring does | → | Tests that fail intermittently are retried until green | ✗ non | discuter |
| Every team has its own logging format | → | The on-call rota has the same two people on it most weeks | ✗ non | discuter |
| Every team has its own logging format | → | Post-mortems assign actions to people who were not in the room | ✗ non | discuter |
| Every team has its own logging format | → | Post-mortem actions are written up and then never scheduled | ✗ non | discuter |
| Post-mortem actions are written up and then never scheduled | → | Tests that fail intermittently are retried until green | ✗ non | discuter |
| The on-call rota has the same two people on it most weeks | → | Error budgets exist on a slide and nowhere else | ✗ non | discuter |
| Alerts fire so often that on-call mutes them | → | The on-call rota has the same two people on it most weeks | ✗ non | discuter |
| One engineer knows how the payment path actually works | → | Tests that fail intermittently are retried until green | ✗ non | discuter |
| Tests that fail intermittently are retried until green | → | Post-mortems assign actions to people who were not in the room | ✗ non | discuter |
| Feature work always outranks reliability work in planning | → | Alerts fire so often that on-call mutes them | ✗ non | discuter |
| Customers report outages before monitoring does | → | Feature work always outranks reliability work in planning | ✗ non | discuter |
| One engineer knows how the payment path actually works | → | Customers report outages before monitoring does | ✗ non | discuter |
| Post-mortems assign actions to people who were not in the room | → | Customers report outages before monitoring does | ✗ non | discuter |
| Error budgets exist on a slide and nowhere else | → | Feature work always outranks reliability work in planning | ✗ non | discuter |
| Alerts fire so often that on-call mutes them | → | Post-mortems assign actions to people who were not in the room | ✗ non | discuter |
| One engineer knows how the payment path actually works | → | Post-mortem actions are written up and then never scheduled | ✗ non | discuter |
| The on-call rota has the same two people on it most weeks | → | One engineer knows how the payment path actually works | ✗ non | discuter |
| Tests that fail intermittently are retried until green | → | One engineer knows how the payment path actually works | ✗ non | discuter |
| The on-call rota has the same two people on it most weeks | → | Every team has its own logging format | ✗ non | discuter |
| Alerts fire so often that on-call mutes them | → | Tests that fail intermittently are retried until green | ✗ non | discuter |
| Every team has its own logging format | → | One engineer knows how the payment path actually works | ✗ non | discuter |
| The on-call rota has the same two people on it most weeks | → | Post-mortems assign actions to people who were not in the room | ✗ non | discuter |
| Post-mortems assign actions to people who were not in the room | → | Tests that fail intermittently are retried until green | ✗ non | discuter |
| Every team has its own logging format | → | Alerts fire so often that on-call mutes them | ✗ non | discuter |
| The on-call rota has the same two people on it most weeks | → | Feature work always outranks reliability work in planning | ✗ non | discuter |
| Tests that fail intermittently are retried until green | → | Feature work always outranks reliability work in planning | ✗ non | discuter |
| Tests that fail intermittently are retried until green | → | Post-mortem actions are written up and then never scheduled | ✗ non | discuter |
| Error budgets exist on a slide and nowhere else | → | Alerts fire so often that on-call mutes them | ✗ non | discuter |
| Error budgets exist on a slide and nowhere else | → | Post-mortem actions are written up and then never scheduled | ✗ non | discuter |
| Post-mortems assign actions to people who were not in the room | → | Feature work always outranks reliability work in planning | ✗ non | discuter |
| One engineer knows how the payment path actually works | → | Feature work always outranks reliability work in planning | ✗ non | discuter |
| Error budgets exist on a slide and nowhere else | → | Every team has its own logging format | ✗ non | discuter |
| Tests that fail intermittently are retried until green | → | Error budgets exist on a slide and nowhere else | ✗ non | discuter |
| Post-mortem actions are written up and then never scheduled | → | Error budgets exist on a slide and nowhere else | ✗ non | discuter |
| Post-mortem actions are written up and then never scheduled | → | Customers report outages before monitoring does | ✗ non | discuter |
| Every team has its own logging format | → | Tests that fail intermittently are retried until green | ✗ non | discuter |
| One engineer knows how the payment path actually works | → | Post-mortems assign actions to people who were not in the room | ✗ non | discuter |
| Feature work always outranks reliability work in planning | → | One engineer knows how the payment path actually works | ✗ non | discuter |
| Every team has its own logging format | → | Feature work always outranks reliability work in planning | ✗ non | discuter |
| One engineer knows how the payment path actually works | → | Error budgets exist on a slide and nowhere else | ✗ non | discuter |
| Alerts fire so often that on-call mutes them | → | Error budgets exist on a slide and nowhere else | ✗ non | discuter |
| Post-mortems assign actions to people who were not in the room | → | One engineer knows how the payment path actually works | ✗ non | discuter |
| Alerts fire so often that on-call mutes them | → | Feature work always outranks reliability work in planning | ✗ non | discuter |
| One engineer knows how the payment path actually works | → | Every team has its own logging format | ✗ non | discuter |
| Feature work always outranks reliability work in planning | → | The on-call rota has the same two people on it most weeks | ✗ non | discuter |
| Post-mortem actions are written up and then never scheduled | → | Feature work always outranks reliability work in planning | ✗ non | discuter |
| The same three services cause most of the pages | → | The service map in the wiki is two years old | ✗ non | discuter |
| Rollbacks take longer than the outage they are meant to end | → | The same three services cause most of the pages | ✗ non | discuter |
| Config lives in five places and drifts between them | → | Rollbacks take longer than the outage they are meant to end | ✗ non | discuter |
| … | → | The staging environment does not resemble production | ✗ non | discuter |
| The incident channel fills with people asking for status instead of giving it | → | Config lives in five places and drifts between them | ✗ non | discuter |
| Reliability has no budget line of its own | → | Retries are unbounded, so a slow dependency becomes a flood | ✗ non | discuter |
| The incident channel fills with people asking for status instead of giving it | → | Rollbacks take longer than the outage they are meant to end | ✗ non | discuter |
| The same three services cause most of the pages | → | The incident channel fills with people asking for status instead of giving it | ✗ non | discuter |
| The service map in the wiki is two years old | → | Retries are unbounded, so a slow dependency becomes a flood | ✗ non | discuter |
| Config lives in five places and drifts between them | → | Feature flags are never cleaned up, so nobody knows which paths are live | ✗ non | discuter |
| The incident channel fills with people asking for status instead of giving it | → | Feature flags are never cleaned up, so nobody knows which paths are live | ✗ non | discuter |
| Retries are unbounded, so a slow dependency becomes a flood | → | The service map in the wiki is two years old | ✗ non | discuter |
| … | → | Reliability has no budget line of its own | ✗ non | discuter |
| The service map in the wiki is two years old | → | Rollbacks take longer than the outage they are meant to end | ✗ non | discuter |
| Feature flags are never cleaned up, so nobody knows which paths are live | → | Reliability has no budget line of its own | ✗ non | discuter |
| Feature flags are never cleaned up, so nobody knows which paths are live | → | The incident channel fills with people asking for status instead of giving it | ✗ non | discuter |
| Rollbacks take longer than the outage they are meant to end | → | … | ✗ non | discuter |
| The incident channel fills with people asking for status instead of giving it | → | Reliability has no budget line of its own | ✗ non | discuter |
| Config lives in five places and drifts between them | → | The staging environment does not resemble production | ✗ non | discuter |
| … | → | Config lives in five places and drifts between them | ✗ non | discuter |
| Feature flags are never cleaned up, so nobody knows which paths are live | → | … | ✗ non | discuter |
| Reliability has no budget line of its own | → | Config lives in five places and drifts between them | ✗ non | discuter |
| The incident channel fills with people asking for status instead of giving it | → | Retries are unbounded, so a slow dependency becomes a flood | ✗ non | discuter |
| Rollbacks take longer than the outage they are meant to end | → | The incident channel fills with people asking for status instead of giving it | ✗ non | discuter |
| The same three services cause most of the pages | → | Reliability has no budget line of its own | ✗ non | discuter |
| … | → | Feature flags are never cleaned up, so nobody knows which paths are live | ✗ non | discuter |
| Feature flags are never cleaned up, so nobody knows which paths are live | → | The service map in the wiki is two years old | ✗ non | discuter |
| Config lives in five places and drifts between them | → | Reliability has no budget line of its own | ✗ non | discuter |
| The staging environment does not resemble production | → | Feature flags are never cleaned up, so nobody knows which paths are live | ✗ non | discuter |
| Rollbacks take longer than the outage they are meant to end | → | Feature flags are never cleaned up, so nobody knows which paths are live | ✗ non | discuter |
| Retries are unbounded, so a slow dependency becomes a flood | → | Config lives in five places and drifts between them | ✗ non | discuter |
| The service map in the wiki is two years old | → | Reliability has no budget line of its own | ✗ non | discuter |
| The service map in the wiki is two years old | → | Config lives in five places and drifts between them | ✗ non | discuter |
| The staging environment does not resemble production | → | Config lives in five places and drifts between them | ✗ non | discuter |
| Rollbacks take longer than the outage they are meant to end | → | The service map in the wiki is two years old | ✗ non | discuter |
| The staging environment does not resemble production | → | The service map in the wiki is two years old | ✗ non | discuter |
| The staging environment does not resemble production | → | Retries are unbounded, so a slow dependency becomes a flood | ✗ non | discuter |
| … | → | The same three services cause most of the pages | ✗ non | discuter |
| Retries are unbounded, so a slow dependency becomes a flood | → | Feature flags are never cleaned up, so nobody knows which paths are live | ✗ non | au premier plan ↑ · fermer ✕ |
| … | → | The service map in the wiki is two years old | ✗ non | discuter |
| Config lives in five places and drifts between them | → | Retries are unbounded, so a slow dependency becomes a flood | ✗ non | discuter |
| The staging environment does not resemble production | → | … | ✗ non | discuter |
| Reliability has no budget line of its own | → | … | ✗ non | discuter |
| The same three services cause most of the pages | → | Config lives in five places and drifts between them | ✗ non | discuter |
| … | → | The incident channel fills with people asking for status instead of giving it | ✗ non | discuter |
| Feature flags are never cleaned up, so nobody knows which paths are live | → | The staging environment does not resemble production | ✗ non | discuter |
| Retries are unbounded, so a slow dependency becomes a flood | → | The incident channel fills with people asking for status instead of giving it | ✗ non | discuter |
| Config lives in five places and drifts between them | → | … | ✗ non | discuter |
| The staging environment does not resemble production | → | The same three services cause most of the pages | ✗ non | discuter |
| Reliability has no budget line of its own | → | Rollbacks take longer than the outage they are meant to end | ✗ non | discuter |
| Config lives in five places and drifts between them | → | The same three services cause most of the pages | ✗ non | discuter |
| The staging environment does not resemble production | → | The incident channel fills with people asking for status instead of giving it | ✗ non | discuter |
| Rollbacks take longer than the outage they are meant to end | → | The staging environment does not resemble production | ✗ non | discuter |
| … | → | Rollbacks take longer than the outage they are meant to end | ✗ non | discuter |
| Config lives in five places and drifts between them | → | The service map in the wiki is two years old | ✗ non | discuter |
| The same three services cause most of the pages | → | Retries are unbounded, so a slow dependency becomes a flood | ✗ non | discuter |
| Rollbacks take longer than the outage they are meant to end | → | Config lives in five places and drifts between them | ✗ non | discuter |
| Reliability has no budget line of its own | → | The service map in the wiki is two years old | ✗ non | discuter |
| Retries are unbounded, so a slow dependency becomes a flood | → | … | ✗ non | discuter |
| Feature flags are never cleaned up, so nobody knows which paths are live | → | Config lives in five places and drifts between them | ✗ non | discuter |
| Config lives in five places and drifts between them | → | The incident channel fills with people asking for status instead of giving it | ✗ non | discuter |
| Feature flags are never cleaned up, so nobody knows which paths are live | → | Retries are unbounded, so a slow dependency becomes a flood | ✗ non | discuter |
| The service map in the wiki is two years old | → | … | ✗ non | discuter |
| The incident channel fills with people asking for status instead of giving it | → | The same three services cause most of the pages | ✗ non | discuter |
| Rollbacks take longer than the outage they are meant to end | → | Retries are unbounded, so a slow dependency becomes a flood | ✗ non | discuter |
| Reliability has no budget line of its own | → | The same three services cause most of the pages | ✗ non | discuter |
| Retries are unbounded, so a slow dependency becomes a flood | → | Rollbacks take longer than the outage they are meant to end | ✗ non | discuter |
| The same three services cause most of the pages | → | Feature flags are never cleaned up, so nobody knows which paths are live | ✗ non | discuter |
| The service map in the wiki is two years old | → | The incident channel fills with people asking for status instead of giving it | ✗ non | discuter |
| The service map in the wiki is two years old | → | The same three services cause most of the pages | ✗ non | discuter |
| The incident channel fills with people asking for status instead of giving it | → | … | ✗ non | discuter |
| The staging environment does not resemble production | → | Reliability has no budget line of its own | ✗ non | discuter |
| The incident channel fills with people asking for status instead of giving it | → | The staging environment does not resemble production | ✗ non | discuter |
| The same three services cause most of the pages | → | Rollbacks take longer than the outage they are meant to end | ✗ non | discuter |
| … | → | Retries are unbounded, so a slow dependency becomes a flood | ✗ non | discuter |
| Retries are unbounded, so a slow dependency becomes a flood | → | Reliability has no budget line of its own | ✗ non | discuter |