kubernetes/ingress-nginx

Zero downtime upgrade

Open

Aperta il 4 mar 2021

Vedi su GitHub
 (36 commenti) (9 reazioni) (0 assegnatari)Go (15.199 star) (7888 fork)batch import
help wantedkind/supportneeds-triagepriority/important-soon

Descrizione

Docs (https://github.com/kubernetes/ingress-nginx/tree/master/charts/ingress-nginx#upgrading-with-zero-downtime-in-production) are mentioning that by default during an upgrade there is a service interruption caused by pods being restarted, and another article is linked with explanation how one can fix that.

But when looking at values.yml I can see both terminationGracePerios and preStop hook is set correctly in order to drain connections.

https://github.com/kubernetes/ingress-nginx/blob/a7fb791132a0dee285b47d0041f4f9acf1e7cff8/charts/ingress-nginx/values.yaml#L237

https://github.com/kubernetes/ingress-nginx/blob/a7fb791132a0dee285b47d0041f4f9acf1e7cff8/charts/ingress-nginx/values.yaml#L583-L587

Is the documentation a little bit outdated or there is something else that I should set to get near zero downtime upgrades?

Guida contributor

Tech stack
gokubernetesdocker
Dominio
devopsinfrastructurecloud
Tipo issue
documentation
DifficoltàQuanto dovrebbe essere impegnativa per un nuovo contributor.
4
Tempo stimatoTempo stimato per completare e verificare uno scope piccolo.
1-2 days
Stato attivitàQuanto è probabile che la issue sia ancora attiva e reviewable.
stale
ChiarezzaQuanto chiaramente la issue descrive problema, scope e risultato atteso.
clear
Prerequisiti
basic kubernetes knowledgehelm chart configurationunderstanding of pod lifecycle hooks
Adatta ai principiantiQuanto la issue è adatta a contributor alla prima esperienza.
30
Direzione di ricerca
Investigate the helm chart values for terminationGracePeriodSeconds and preStop hook in ingress nginx (see values.yaml lines 237, 583-587). Compare with the linked documentation on zero downtime upgrades. Determine if the documentation is outdated or if additional configuration is needed. Check the referenced article for zero downtime strategies and test the behavior during a rolling update. Consider updating the documentation or values if inconsistencies are found.