YouTube Buffering and Quality Locked? Split googlevideo and Account Domains in Clash (2026 Steps)

1. Symptoms that point to split routing, not “YouTube is down”

Outages happen, yet Clash readers should listen for a sharper signature than a generic error page. The watch shell renders, comments and related videos load, but the player wedges in an endless buffering state while the network graph looks idle. Alternatively, resolution refuses to rise past 720p or 480p despite a membership that should unlock higher tiers, which users often mislabel as a pure quality lock bug when the adaptive engine is simply starved or confused about region. Premium renewal prompts that appear only behind a proxy, or sign-in loops where the account tab succeeds in Chrome but the embedded player still thinks you are logged out, usually trace to account API traffic exiting through a different path than the watch page.

Another loud clue appears in the connection log itself. You might see youtube.com or www.youtube.com on PROXY while long runs of rr*.googlevideo.com or similar media hosts show DIRECT—sometimes because a subscription’s “domestic CDN” rule fires early, sometimes because QUIC-based flows bypass what you thought was a uniform stack. Before you burn an evening rotating upstream nodes, confirm whether the failure tracks with a profile change or resolver switch rather than YouTube’s public status dashboard.

If editing YAML still feels intimidating, skim our subscription import tutorial so you know where provider bundles end and your personal rule section begins. The rest of this article assumes you can append suffix rules without breaking schema validation.

2. Why YouTube fractures when only the homepage is proxied

Google deliberately separates the product shell from fat media pulls. The visible page issues HTML and JSON calls to configure the player; encrypted video segments and audio renditions stream from googlevideo-branded hosts whose certificates, SNI, and peering paths can differ sharply from the marketing hostname. A naive profile that proxies youtube.com but leaves unnamed CDN-like hosts on DIRECT can still strand playback because the adaptive bitrate ladder never receives steady throughput on the path you expect.

Account and device integrity checks add another layer. OAuth-style redirects, cookie sync endpoints, and occasional Play-services adjacent calls need a coherent exit, especially when Google classifies sessions as suspicious when TLS fingerprints jump between regions mid-handshake. A mixed graph that proxies the HTML shell through Tokyo while account APIs resolve as if you were domestic is exactly the pattern that produces weird Premium banners or inconsistent entitlement states—not necessarily malicious intervention, just contradictory network signals.

Static blog posts that publish a “full Google suffix list” decay quickly because CDNs and experiments shift. Treat any sample list in this article—and anything you read in forums—as a hypothesis to verify against your own Mihomo captures on the day you troubleshoot in 2026, not as scripture.

3. A checklist before you blame the exit node

Swapping nodes feels decisive; it often wastes time when chrome.exe or the YouTube app never hit Clash in the first place. Work through this sequence while keeping the live connection view open.

1. Decide whether you rely on system proxy or TUN, then confirm the browser or app actually inherits that path (service workers, helper processes, and QUIC sessions can behave differently).
2. Reproduce YouTube buffering, then read the policy column per hostname. Stray DIRECT rows next to googlevideo-related names are the usual smoking gun.
3. Audit DNS: upstream reachability, fake-ip expectations, and whether campus or private DNS rewrites Google-owned names oddly.
4. Expand Clash rules so APIs, static assets, googlevideo edges, and account hosts you truly observe share a coherent policy—not only the apex youtube.com label.
5. Only after routing is coherent, pick stable exits for long sessions and avoid hyperactive failover that reconnects during player init.

For port collisions, invalid YAML, and core startup errors, keep the general Clash troubleshooting guide nearby. Here we focus on streaming surfaces where a single missing suffix masquerades as a platform outage.

4. System proxy versus TUN for browsers and stubborn apps

System proxy remains the lightest option when Chromium or Safari inherits OS settings and nothing else fights them. The familiar failure mode returns: the document request succeeds, yet secondary hosts—sometimes QUIC-first—bypass the proxy, starving the segment pipeline even while the UI looks loaded.

TUN mode lowers the chance that executables silently skirt Clash by pushing routing deeper into the stack. The trade-off is operational friction: permissions, route tables, and occasional clashes with other VPN-class products. If you already followed our TUN mode guide, repeat the experiment while filtering specifically for googlevideo substrings in the log. TUN is not mandatory for everyone; it is the right lever when evidence shows stubborn bypass despite apparently correct YAML.

Televisions and embedded platforms may ignore PC proxies entirely—compare notes with our OpenWrt side-router guide if you need gateway-level routing for a living-room client. This article emphasizes desktop and phone browsers plus official apps where system modes or TUN apply.

5. DNS, fake-ip, and why streaming feels fragile

Clash’s fake-ip mode answers quickly with synthetic addresses, yet it tightly couples DNS to rule evaluation. When the resolver and the rule engine disagree about what a media hostname “means,” you may see TLS retries, half-open HTTP/2 sessions, and players stuck in buffering while the UI still shows optimistic thumbnails.

Practical mitigation usually has two parts. First, ensure your upstream resolvers are reachable through the policy path you expect for general browsing, and avoid chains that intermittently drop international queries. Second, consider targeted policies—commonly nameserver-policy in Mihomo-compatible cores—for suffixes that dominate YouTube captures, such as youtube.com, googlevideo.com, ytimg.com, or ggpht.com. Always verify keys against the documentation bundled with your exact core build instead of copying aged snippets.

When DNS fixes clear most symptoms without changing proxy groups, you have strong evidence the bottleneck was resolution, not bandwidth. That distinction tells you whether to invest in resolver hygiene or node stability next.

6. Adaptive bitrate, “quality lock,” and throughput illusions

YouTube’s player does not pick arbitrary caps; it probes available throughput along multiple renditions. If segments arrive late because the googlevideo path is slow, lossy, or inconsistent with the control-plane exit—classic partial routing—the ladder ratchets down and the gear menu looks like a hard quality lock even when a speed test elsewhere shows hundreds of megabits.

Another subtle pattern appears when IPv6 paths disagree with IPv4. If one family rides your tunnel while the other leaks around it, QUIC may prefer the “wrong” interface, producing jitter that the adaptive engine reads as congestion. During triage, note whether disabling QUIC or preferring IPv4 in the client—only as an experiment—changes segment behavior, then return to defaults once you understand the interaction.

Premium-specific features add entitlement checks on top of raw throughput. If account APIs still think you are unsigned-in because accounts.google.com traffic follows a different resolver or policy than the player shell, you may see restricted behaviors that resemble censorship bugs. Align account-facing suffixes with the same coherent policy unless you have a deliberate reason to split them—and document that choice.

7. How to collect hostnames you can defend in a ticket

Open Chromium devtools on the watch page, enable preserve log, filter by domain, and start playback that reproduces the stall. Note document requests, XHR or fetch calls for player configuration, adaptive manifest traffic, and the heavy byte flows that land on googlevideo hosts. Pay attention to Web Workers or service workers that may issue parallel connections your first glance missed.

For mobile apps where devtools are awkward, pair the platform’s per-app network view with the Mihomo table: if a hostname appears in the OS network log but never in Clash, you still have a visibility problem. On Windows, filtering the live view for googlevideo, ytimg, and ggpht substrings usually surfaces critical rows within seconds.

Timestamp your findings. CDNs shift; a list titled “works April 2025” is a liability by 2026. Future you will thank present you when an experiment flips overnight.

8. Domain buckets from shell APIs to googlevideo edges

After collection, group hosts so your configuration stays readable. Names drift; verify each row against your own logs before pasting into production YAML.

The mental model parallels our Netflix geo and CDN split routing article—except YouTube’s edge graph emphasizes interactive watch pages, shorter segment cadence, and pervasive Google account coupling rather than long-form episode DRM windows alone.

9. Rule snippets: explicit coverage and clean ordering

The YAML fragments below illustrate steering traffic to a proxy group named PROXY. Rename that token to match your policy label and insert these lines before broad provider rules that might prematurely return DIRECT for “domestic” CDNs that Google also uses.

# Example only — replace PROXY; verify suffixes against your Mihomo logs
rules:
  - DOMAIN-SUFFIX,youtube.com,PROXY
  - DOMAIN-SUFFIX,googlevideo.com,PROXY
  - DOMAIN-SUFFIX,ytimg.com,PROXY
  - DOMAIN-SUFFIX,ggpht.com,PROXY
  - DOMAIN-SUFFIX,googleapis.com,PROXY
  - DOMAIN-SUFFIX,gstatic.com,PROXY
  - DOMAIN-SUFFIX,accounts.google.com,PROXY

Expand with additional DOMAIN-SUFFIX rows for Google properties your captures prove necessary—never paste mystery keywords because a blog told you to. Prefer explicit suffixes over shotgun DOMAIN-KEYWORD rules unless you must; substring matches are powerful and easy to overfit across unrelated traffic.

If your subscription injects aggressive geo rules, duplicate critical YouTube-related lines in a user-controlled section that wins on precedence, or merge providers thoughtfully so your exceptions survive merges from remote updates.

10. Mobile apps, background playback, and OS-level gotchas

Phones introduce battery-aware scheduling, per-app cellular toggles, private DNS profiles, and “restricted networking” states that desktops rarely surface. If YouTube works on Wi-Fi at home but stutters on cellular—or only after you reopen the app—check whether the client is allowed to use mobile data and whether an always-on VPN slot conflicts with Clash’s TUN.

Picture-in-picture and background audio may keep auxiliary connections alive; read the log during those transitions to see if policy flips coincide with stalls. Offline downloads should not need live googlevideo edges, yet license pings may still call home—compare failing streams with offline playback to isolate live-only names.

11. GUI workflow: logs are the source of truth

Desktop clients such as Clash Verge Rev expose live connections, DNS panes, and rule editors side by side. When YouTube misbehaves, filter for googlevideo and read the chosen policy per row. If anything sensitive shows DIRECT while sibling hosts use PROXY, fix precedence before swapping servers.

If the baseline GUI still feels unfamiliar, walk through the Clash Verge Rev setup guide to confirm ports, subscriptions, and first launch before you chase YouTube-specific ghosts.

12. How this differs from Spotify or Netflix guides

Our Spotify split article stresses short audio segments and client configuration loops; Netflix streaming focuses on episodic video and sustained CDN throughput with heavy DRM flavor. YouTube combines a Google-wide account plane, aggressive QUIC usage, and a huge googlevideo fan-out—copying only one sibling guide verbatim often leaves you half-proxied.

Enterprise networks deserve a reminder: TLS inspection and split-horizon DNS can break Google surfaces even when Clash YAML is pristine. If only YouTube-facing traces fail while unrelated HTTPS succeeds, involve the network team with connection logs rather than assuming your exit node is at fault.

13. Terms, ethics, and what this article is not

Google’s terms of service and local laws govern what you may do with YouTube products. This article describes network hygiene for viewers whose routing is accidentally fractured—travelers, students on constrained campus networks, and power users who see YouTube buffering because of configuration—not a recipe to bypass regional licensing you are not entitled to, nor an endorsement of credential sharing.

When you want to read upstream source or file issues, community repositories linked from our docs remain appropriate. Keep GitHub distinct from everyday install paths: readers should fetch maintained desktop builds from this site’s download flow rather than chasing raw release assets in threads.

14. Close with evidence, not superstition

Endless buffering and mysterious quality lock reports are maddening because the YouTube UI still looks authoritative even when the network path is fractured. Treat every spinner as a prompt to open the connection log, read policies row by row, and reconcile DNS with the hostnames your client requested. Coherent Clash split routing that covers googlevideo alongside YouTube shell and Google account traffic—not only the homepage—is the mechanical layer; calm, stable nodes are the polish once the path is honest. For 2026 viewers juggling browsers, native apps, and household gateways, that discipline matters more than cargo-cult keyword lists.

Compared with toggling random VPNs, a maintained desktop client with Mihomo integration keeps diagnostics visible and tames YAML foot-guns when Google shifts edges quietly in the background. → Download Clash for free and experience the difference