Social Wire
Tweeted
Nov 3, 2026
VEXXHOST, Inc.
@vexxhost Operators: when you drain a compute node for maintenance, does every instance live-migrate, or do some get a cold migration or a planned stop? Curious where people draw that line.
VEXXHOST, Inc.
@vexxhost Operators: when you drain a compute node for maintenance, does every instance live-migrate, or do some get a cold migration or a planned stop? Curious where people draw that line.
Tweeted
Nov 3, 2026
VEXXHOST, Inc.
@vexxhost Both cost models step by step, with break-even volume and what N+1 headroom does to it: vexxhost.com/blog/reducing-…
VEXXHOST, Inc.
@vexxhost Both cost models step by step, with break-even volume and what N+1 headroom does to it: vexxhost.com/blog/reducing-…
Tweeted
Nov 3, 2026
VEXXHOST, Inc.
@vexxhost A dedicated inference node sized for peak can cost more per result than an API at average demand. In an illustrative model, a $25/hour node completes 3,600 successful outcomes an hour at the target p95, or $0.0069 each. At the average of 833 an hour, it comes to $0.030.
VEXXHOST, Inc.
@vexxhost A dedicated inference node sized for peak can cost more per result than an API at average demand. In an illustrative model, a $25/hour node completes 3,600 successful outcomes an hour at the target p95, or $0.0069 each. At the average of 833 an hour, it comes to $0.030.
Tweeted
Nov 3, 2026
VEXXHOST, Inc.
@vexxhost The full baseline also covers Neutron and Barbican: vexxhost.com/blog/building-…
VEXXHOST, Inc.
@vexxhost The full baseline also covers Neutron and Barbican: vexxhost.com/blog/building-…
Tweeted
Nov 3, 2026
VEXXHOST, Inc.
@vexxhost If you enable lockout in [security_compliance], exempt service users first: openstack user set --ignore-lockout-failure-attempts <service-user-id> A locked-out service user stops its service.
VEXXHOST, Inc.
@vexxhost If you enable lockout in [security_compliance], exempt service users first: openstack user set --ignore-lockout-failure-attempts <service-user-id> A locked-out service user stops its service.
Tweeted
Nov 3, 2026
VEXXHOST, Inc.
@vexxhost On a multi-node Keystone, rotate on one node with keystone-manage fernet_rotate, then distribute that key repository to every other node before the next rotation. The staged key (key 0) can decrypt tokens but never creates them, which covers the gap.
VEXXHOST, Inc.
@vexxhost On a multi-node Keystone, rotate on one node with keystone-manage fernet_rotate, then distribute that key repository to every other node before the next rotation. The staged key (key 0) can decrypt tokens but never creates them, which covers the gap.
Tweeted
Nov 3, 2026
VEXXHOST, Inc.
@vexxhost Keystone's Fernet docs give a sizing rule for key rotation: max_active_keys = (token_expiration / rotation_frequency) + 2 Both in hours. With 24-hour tokens rotated every 6 hours, that's 6. Set it lower and Keystone can prune a key that still validates unexpired tokens.
VEXXHOST, Inc.
@vexxhost Keystone's Fernet docs give a sizing rule for key rotation: max_active_keys = (token_expiration / rotation_frequency) + 2 Both in hours. With 24-hour tokens rotated every 6 hours, that's 6. Set it lower and Keystone can prune a key that still validates unexpired tokens.
Tweeted
Nov 3, 2026
VEXXHOST, Inc.
@vexxhost Why 21 days: cert-manager renews a 90-day certificate with 30 days left, so the first alert means renewal has stalled for over a week. Runbook for both alerts: vexxhost.github.io/atmosphere/adm…
VEXXHOST, Inc.
@vexxhost Why 21 days: cert-manager renews a 90-day certificate with 30 days left, so the first alert means renewal has stalled for over a week. Runbook for both alerts: vexxhost.github.io/atmosphere/adm…
Tweeted
Nov 3, 2026
VEXXHOST, Inc.
@vexxhost Atmosphere 7.11 adds alerts for TLS certificates that NGINX Ingress serves from a Kubernetes Secret. NginxIngressTLSCertificateExpiringSoon fires under 21 days and NginxIngressTLSCertificateExpiringCritical pages under 7. One alert per Secret, so a wildcard cert raises one.
VEXXHOST, Inc.
@vexxhost Atmosphere 7.11 adds alerts for TLS certificates that NGINX Ingress serves from a Kubernetes Secret. NginxIngressTLSCertificateExpiringSoon fires under 21 days and NginxIngressTLSCertificateExpiringCritical pages under 7. One alert per Secret, so a wildcard cert raises one.
Tweeted
Nov 3, 2026
VEXXHOST, Inc.
@vexxhost Backend choice and acceleration setup: vexxhost.github.io/atmosphere/dep…
VEXXHOST, Inc.
@vexxhost Backend choice and acceleration setup: vexxhost.github.io/atmosphere/dep…