Issues

provider: manifest manager trusts the last DeploymentUpdated event over a fresh chain read, so a missed event rejects every manifest push with 'manifest version validation failed'

#662

· baktun14 opened

on Sep 23, 2026

100% single-core CPU busy loops.

awaiting-triage

#661

· dl2klab opened

on Sep 22, 2026

akash-node Helm chart: only 4 of 50 default persistent peers are alive, and there is no seeds option — nodes can't recover peers after an outage (meta.json: 7/55, its only seed dead)

#660

· cmartins88 opened

on Sep 19, 2026

Provider hardening: nothing limits how fast a tenant pod opens outbound connections — one lease's HTTP flood exhausted a provider's NAT and took it offline (proposed node-level guard + owner deny-list)

#659

· cmartins88 opened

on Sep 19, 2026

provider: gorilla/context entries are never cleared - every REST request permanently retains its *http.Request (unbounded memory growth to OOM)

#658

· cmartins88 opened

on Sep 9, 2026

kill_zombie_parents.sh kills tenant workloads: it SIGTERM/SIGKILLs a container's PID 1

#657

· cmartins88 opened

on Sep 6, 2026

Homenode - unable to complete onboarding due to tax form not loading

#656

· anomaly256 opened

on Sep 3, 2026

provider: reservations for already-running deployments start unallocated, double-counting the entire running estate at boot

#654

· cmartins88 opened

on Aug 27, 2026

provider: ClusterDeploymentUpdated permanently un-allocates a reservation, shrinking advertised capacity until restart

#653

· cmartins88 opened

on Aug 27, 2026

operator-inventory: hardware-discovery pod is never recreated after a node drain (worker stops watching its pod once Running)

#652

· cmartins88 opened

on Aug 20, 2026

Provider: public bid-policy endpoint - let providers declare what they will not bid on

#648

· cmartins88 opened

on Jul 28, 2026

[docs] Missing troubleshooting guidance: /v1/inventory returns empty storage array when StorageClass label akash.network=true is not present

#645

· jjozzietech opened

on Jul 16, 2026

Navigation Menu

All issues

provider: manifest manager trusts the last DeploymentUpdated event over a fresh chain read, so a missed event rejects every manifest push with 'manifest version validation failed'

100% single-core CPU busy loops.

akash-node Helm chart: only 4 of 50 default persistent peers are alive, and there is no seeds option — nodes can't recover peers after an outage (meta.json: 7/55, its only seed dead)

Provider hardening: nothing limits how fast a tenant pod opens outbound connections — one lease's HTTP flood exhausted a provider's NAT and took it offline (proposed node-level guard + owner deny-list)

provider: gorilla/context entries are never cleared - every REST request permanently retains its *http.Request (unbounded memory growth to OOM)

kill_zombie_parents.sh kills tenant workloads: it SIGTERM/SIGKILLs a container's PID 1

Homenode - unable to complete onboarding due to tax form not loading

provider: reservations for already-running deployments start unallocated, double-counting the entire running estate at boot

provider: ClusterDeploymentUpdated permanently un-allocates a reservation, shrinking advertised capacity until restart

operator-inventory: hardware-discovery pod is never recreated after a node drain (worker stops watching its pod once Running)

Provider: public bid-policy endpoint - let providers declare what they will not bid on

[docs] Missing troubleshooting guidance: /v1/inventory returns empty storage array when StorageClass label akash.network=true is not present

Search results