Prisma Access egress and public IP addresses can change as a result of autoscaling or infrastructure upgrades, so any allow-list dependent on those addresses (SaaS tenant restrictions, partner firewalls, third-party services) needs a reliable way to stay current. Palo Alto Networks addresses this with two complementary mechanisms. First, an Egress IP Notification URL — the webhook referenced in option A — can be configured under Infrastructure Settings so that Prisma Access sends an HTTP POST a few seconds before a new IP address becomes active, giving downstream automation advance warning to update firewall or SaaS allow-lists before the change takes effect. Second, retrieving the actual address list requires authenticating to the Egress/Public IP retrieval API using an API key that is generated and copied from the service infrastructure settings, as described in option B; this key is passed in the request header when calling the retrieval endpoint. There is no separate " enable the Egress IP API endpoint " toggle, since the retrieval API is available by default once a key is generated — making option C incorrect. Authentication to this API is strictly key-based, not certificate-based, so downloading a client certificate (option D) is not a supported or required step. Together, the webhook and API key form the complete automation loop: notify, then retrieve and apply.

Forcing all application access through PAB for unmanaged or non-compliant devices requires addressing two distinct application authentication patterns separately, since a single enforcement mechanism cannot cover both. For applications that are integrated with the organization ' s identity provider and support SSO, the Enforce SSO setting ensures that any attempt to authenticate to that application is redirected specifically through the PAB-brokered session rather than allowing a direct, out-of-band login that would bypass PAB ' s controls entirely — this closes the most common bypass path for SSO-capable SaaS and web applications. For applications that are not SSO-enabled and therefore cannot be gated the same way, Account Protection provides the complementary enforcement mechanism, restricting direct credential-based access to those applications outside of the PAB session so that even non-SSO applications cannot be reached through an unmanaged, unenforced path. Together, these two settings comprehensively cover both application authentication models, which is why options A and B form the correct pair. The PAB Extension (option C) is a deployment and traffic-redirection mechanism, but on its own it does not enforce that access only happens through PAB — a user could still, without SSO enforcement or Account Protection in place, log in to an application directly outside the extension ' s redirected session. Device Posture (option D) is used to assess and act on a device ' s compliance state to allow or block traffic generally; it is a conditional access input, not the specific enforcement mechanism that closes the SSO and non-SSO application bypass paths described in the question.

Certificate pinning is a well-documented, expected source of SSL decryption failures on any inline TLS proxy, including the Prisma Access decryption engine. When an application (in this case, one interacting with google.com endpoints) has pinned the exact certificate or public key it expects from the origin server, it will reject the substitute certificate that Prisma Access presents during man-in-the-middle SSL Forward Proxy decryption, even though that substitute certificate is validly signed by the organization ' s trusted forward-trust CA. This produces the decrypt error log entries referencing the failed hostname, and the server-side certificate pinning behavior is the root cause described in option C. Because pinning cannot be bypassed by adjusting client trust stores or firewall decryption profiles, the only supported remediation is a policy-based exception: creating a Do Not Decrypt rule scoped to the affected hostname, google.com in this scenario, so that traffic to that specific destination bypasses SSL decryption entirely and the application ' s pinning check succeeds against the real origin certificate. Client misconfiguration (option A) is not supported by log entries that clearly attribute the failure to certificate validation against a known-pinning application. The certificates.godaddy.com reference in the log is incidental to the underlying trust chain being validated, not the actual site the user is browsing to, so a decrypt exclusion should be scoped to google.com, not to the CA hostname, making option D incorrect.

A burst logon event, where a large branch office population authenticates and begins generating DNS lookups within a narrow five-minute window, is exactly the scenario Prisma Access ' s DNS proxy sizing limits and caching behavior are designed to withstand, and understanding those documented defaults explains user-visible behavior during onboarding rushes like this one. The DNS proxy on Prisma Access caches every resolved record it handles, not merely a curated subset of " frequently used " hostnames, for a fixed default duration of 300 seconds; this blanket caching (option D) is precisely what allows a large burst of simultaneous, repeated lookups for the same common destinations (SaaS portals, internal domains, update servers) to be served from cache rather than generating a fresh upstream query for every single request, which is critical to sustaining performance during a synchronized-logon event. The DNS proxy also has a defined ceiling on how many TCP-based DNS requests it will hold pending concurrently, documented as 64 (option B); in a large burst scenario this is the throttling limit that governs how much simultaneous TCP DNS load the proxy will queue before applying back-pressure. Option A misstates the caching behavior — caching is not selective to " frequently used " hostnames, it applies broadly for the TTL period. Option C references a retry count that is not the documented, relevant limiting factor in this burst-capacity scenario.

App Acceleration works by having Prisma Access decrypt, optimize, and re-encrypt SaaS application traffic across its backbone to reduce round-trip latency and improve throughput to well-known, high-volume SaaS destinations, and that optimization is fundamentally dependent on SSL Forward Proxy decryption already being functional and trusted end-to-end. Two certificate-related prerequisites make this possible: a Forward Trust Certificate configured for SSL decryption, which Prisma Access presents to the client in place of the SaaS provider ' s original certificate when it performs the man-in-the-middle decryption necessary to inspect and accelerate the session, and that certificate ' s issuing CA must be distributed to and trusted by client endpoints as a Trusted Root CA, so that browsers and applications do not throw certificate warnings or reject the substituted certificate. Both of these are explicit, documented prerequisites for App Acceleration to function correctly, which makes options C and D the correct pair. There is no dedicated " acceleration agent " software component that must be installed on client machines (option A); App Acceleration operates transparently at the Prisma Access infrastructure level for tunneled or proxied users, not through an endpoint agent add-on. QoS (option B) is a separate traffic-shaping capability used to prioritize bandwidth for specific application classes; it is not a prerequisite for App Acceleration to be enabled and is functionally unrelated to the decryption trust chain that acceleration depends on.

The branch requirement is internet filtering plus data center connectivity, which means every branch location needs Prisma Access to become its default gateway to the internet while still exchanging specific internal routes with the data center. Enabling eBGP on the Remote Networks connection is the scalable way to accomplish this: dynamic routing lets the CPE and Prisma Access exchange branch subnet reachability automatically, without the administrative burden of manually maintaining static routes across every site as the branch network changes — critical for a multi-branch B2B/enterprise deployment. Enabling the Advertise Default Route option on the Remote Networks connection is what actually delivers the internet-filtering requirement: it causes Prisma Access to advertise a 0.0.0.0/0 route to the branch CPE over the tunnel so that all branch-originated internet-bound traffic is pulled into Prisma Access for inspection, rather than breaking out locally. Static routes (options A and C) are technically workable at very small scale, but they do not scale for multi-site deployments, are error-prone to maintain, and do nothing on their own to steer default (internet) traffic into the tunnel the way the Advertise Default Route setting does. eBGP with Advertise Default Route is the documented best-practice combination for branch internet filtering and site connectivity through Remote Networks.

The reported symptom — traffic intermittently falling through to the bottom Catch-All Deny rule, bypassing the HIP-based policy that should be matching first, and being resolved simply by refreshing the VPN connection — is a classic signature of stale or lost user-to-IP mapping combined with expired HIP state, rather than a fundamental policy configuration error, which is why refreshing the session (forcing re-authentication and a fresh HIP report) restores correct behavior. If user mapping for the connected session is being learned or refreshed from a source other than the gateway ' s own authentication event (for example, User-ID redistribution or another mapping source with different timing or reliability characteristics than the gateway ' s native session state), that mapping can become inconsistent with the live GlobalProtect session, causing the HIP-enforced rule ' s user-based match criteria to intermittently fail — this is option B. Separately, the firewall periodically expects HIP report checks from the connected endpoint to keep its HIP-based match state current; if a report check is missed due to a client-side timing issue or transient connectivity blip, the firewall can lose the HIP match state for that session even though the tunnel itself remains up, causing subsequent traffic to fail HIP-based rule matching and fall through to the deny-all rule — this is option C. " Collect HIP data " not being enabled (option A) would cause a total, consistent failure to match HIP-based policy from the outset, not the intermittent pattern described. A time-of-day schedule on the HIP rule (option D) would produce a predictable, not intermittent and refresh-resolved, pattern of denial.