Handling Traffic Spikes with Varnish Grace Mode Magento 2

1. TTL and Grace Setup

When Varnish caches an object, it defines two values in the vcl_backend_response subroutine:

• TTL (Time to Live): It is the period during which you consider cached content as fresh. The cached content is valid for 5 minutes.

• Grace Period: The duration during which you allow Varnish to serve the content. It is after the TTL expires that the Magento backend is slow or fails. After expiry, you can still serve stale content for up to 10 minutes if needed.

2. During Normal Operation

When a user visits a page:

• If the cached object is still within its TTL period, Varnish serves it.

• There is no contact with the backend, ensuring ultra-fast response times.

3. When TTL Expires

If the TTL has expired:

• Varnish attempts to fetch a fresh copy from the Magento backend. It considers the content stale.

• If the backend is slow and takes longer than expected. Varnish serves the expired content from the cache immediately to the user. It ensures no delay.

• Meanwhile, it fetches the updated version in the background and refreshes the cache.

4. If Backend Fails

If the backend server is completely down:

• Varnish checks if the cached content is still within the grace window.

• If yes, it delivers the stale content as a fallback response.

• It avoids blank pages or 500 errors for end users.

• Once the grace period expires, the content is no longer served. It shows an error unless another fallback mechanism is in place.

6 Important Considerations for Varnish Grace Mode

1. Stale Content Risks

Grace mode helps serve stale content when the Magento backend is slow or unresponsive. While this keeps your site live, it also means:

• Outdated product prices or stock availability.

• For high-frequency inventory updates or flash sales. It could lead to customer dissatisfaction or overselling.

Keep grace durations short for pages with time-sensitive data. Exclude such pages from caching altogether using cacheable.xml and custom cache rules.

2. Handling User-Specific Data

• Magento 2 includes dynamic elements like:

  1. Shopping cart totals

  2. Recently viewed items

  3. Customer login status

  4. Wishlists and account links

• You should not cache them as part of the full page. Use ESI to cache the static page shell while fetching personalized fragments.

• It allows Grace Mode to serve stale content. They don't expose incorrect or private information.

3. Grace Period Duration

Choose the right grace duration:

• Too short, 30s might not help much during real outages.

• Too long, 30m, increases the risk of users seeing outdated content.

Recommended Range:

• 2–5 minutes for most catalog and product pages.

• Lower for dynamic landing pages or promotional banners.

Track how long your backend takes to recover from high loads or minor failures. Set a grace period.

4. Magento Full Page Cache Compatibility

Magento 2 uses its Full Page Cache system that integrates with Varnish. To ensure Grace Mode works:

• Use Magento’s built-in bin/magento varnish:vcl:generate command to create your base VCL.

• Ensure that you honor cache tags to enable selective purging.

• You should not cache pages like the cart and customer/account.

Magento sends headers like Cache-Control and X-Magento-Tags. Configure Varnish to respect and parse these.

5. Test Failover Scenarios

To trust the Grace mode, you must test it under failure conditions:

• Simulate backend slowness, using sleep() in PHP or stopping PHP-FPM.

• Observe whether Varnish delivers stale content within the grace window.

• Ensure that your monitoring tools reflect accurate cache behavior.

Tools for Testing:

• curl \-I or curl \-w %{time\_total} to measure response behavior.

• varnishlog to trace backend fetches and grace hits.

6. Monitoring and Logging

To ensure Grace Mode is working as expected, use:

• varnishlog: See if grace is being used and backend fetch attempts.

• varnishstat: Track real-time statistics, such as cache hits and backend failures.

Consider setting up external monitoring tools, such as Prometheus + Grafana. Set up Varnish Enterprise’s dashboard if using the commercial version.

Create alerts for high backend error rates or spikes in grace responses. It helps detect issues early.

Varnish Grace Mode vs Other Caching Strategies

3 Integration Strategies of Varnish Grace Mode

1. Varnish + Redis Architecture

• Magento 2 supports Redis out of the box for both session and page cache storage. Varnish helps skip the page cache.

• Redis stores volatile session data, such as customer login state and form inputs.

• Varnish handles full-page cache. It uses grace mode to serve stale content if the backend is slow or down.

Benefits

• It reduces pressure on the Magento database by offloading session storage to Redis.

• It enables high availability. Varnish can serve pages, and Redis can provide user-specific info. It is even if Magento degrades.

• It speeds up page delivery and improves fault tolerance.

2. Using Elasticsearch for Search

• Magento 2 uses Elasticsearch as the primary catalog search engine.

• Varnish handles page caching but bypasses search-related traffic. These include layered navigation and live search.

• Magento’s backend or AJAX calls query Elasticsearch.

Benefits

• It ensures a fast and accurate product search experience.

• It separates the search engine load from the Magento backend.

• It keeps catalog pages available during Elasticsearch failures.

3. CDN + Varnish Grace Mode Architecture

• CDN, like Cloudflare and Akamai, is set up in front of Varnish.

• Configure Varnish with grace and keep settings:

• Magento origin servers handle backend processing and uncached requests.

Benefits

• It enables multi-layered caching.

• High availability even when Magento is under stress or unreachable.

• It enables geographic performance optimization. CDN edge nodes cache assets closer to users.

• It handles traffic spikes without straining the backend.

5 Common Challenges and Solutions of Varnish Grace Mode

1. Stale Content Displayed to Users

Varnish may serve outdated product data, including incorrect prices or unavailable items. It is when using grace mode that the backend remains unresponsive.

Solution:

• Set appropriate grace and keep timeouts.

• Use cache tags and ban mechanisms in Magento to invalidate stale content.

• Combine with Edge Side Includes to refresh personalized blocks.

2. Grace Mode Not Triggering as Expected

Varnish might fail to serve stale content if the backend fails. It is especially true if you keep objects that have expired.

Solution:

• Ensure to configure the backend health check in Varnish. Unhealthy backends help trigger grace mode.

• Use set beresp.grace = 1h and set beresp.keep = 6h.

• Use retry logic or failover for backend requests.

3. SEO Risks from Outdated Pages

Search engines may index stale content served during grace periods. It affects SEO and user trust.

Solution:

• Avoid caching pages with the 'noindex' attribute or dynamic, user-specific URLs.

• Use Surrogate-Control headers to fine-tune cache duration per content type.

• Leverage cache tag invalidation in Magento for timely content purges.

4. Difficulty Diagnosing Grace Mode Behavior

Debugging whether the grace mode or the regular cache serves the content. It can be unclear without logging.

Solution:

• Add custom headers in VCL to show cache status.

• Use tools like Varnishlog or New Relic to track the origins of responses.

5. Conflicts with Other Caching Layers

Improper coordination between Varnish Grace Mode and other caching mechanisms. It can cause inconsistent content delivery.

Solution:

• Define cache hierarchy, such as CDN and Varnish.

• Use consistent TTLs and cache headers across layers.

• Invalidate the cache at all layers when critical content updates.

sub vcl_backend_response {
    # Grace Mode Configuration
    # Generated by Varnish Grace Mode Builder
    # Set Time to Live (TTL)
    set beresp.ttl = 5m;
    # Set Grace Period - serve stale content if backend is down
    set beresp.grace = 10m;
    # Set Keep Period - how long to keep in storage
    set beresp.keep = 60m;
    # Content-specific settings
    if (bereq.url ~ "^/catalog/product/") {
        # Product pages
        set beresp.ttl = 5m;
        set beresp.grace = 10m;
    }
    if (bereq.url ~ "^/catalog/category/") {
        # Category pages
        set beresp.ttl = 5m;
        set beresp.grace = 10m;
    }
    if (bereq.url ~ "^/cms/") {
        # CMS pages
        set beresp.ttl = 5m;
        set beresp.grace = 10m;
    }
    # Health check for backend
    if (beresp.status >= 500) {
        # Backend is unhealthy, increase grace
        set beresp.grace = 24h;
    }
}
sub vcl_hit {
    # Handle grace mode logic
    if (obj.ttl >= 0s) {
        # Object is fresh, deliver it
        return (deliver);
    }
    if (obj.ttl + obj.grace > 0s) {
        # Object is in grace period
        if (std.healthy(req.backend_hint)) {
            # Backend is healthy, fetch fresh
            return (miss);
        } else {
            # Backend is down, deliver stale
            return (deliver);
        }
    }
    return (miss);
}

3 Use Cases of Varnish Grace Mode

1. Everlast

• Everlast performs regular Magento patch updates and deploys features late at night.

• During these windows, the backend is in maintenance mode. Varnish still serves stale but rendered HTML pages via grace mode. It helps avoid any user disruption.

• The DevOps team uses custom VCL logic to bypass Magento completely for certain paths. It is during maintenance. It relies on Varnish Grace’s logic to maintain content delivery.

2. JCB Shop

• JCB shop’s eCommerce store encounters sudden traffic surges from international marketing campaigns.

• Grace Mode protects against Magento overload. It is by keeping cached versions available for up to 1 hour. It absorbs massive traffic with no backend pressure.

• Pair Grace mode with fast SSD-based Varnish storage to ensure rapid delivery. Magento’s cache tags help keep high-priority products updated.

3. Graze.com

• Graze.com serves personalized carts and messages via ESI blocks. Varnish caches the shell of product pages and category listings. They pull dynamic blocks in post-load.

• During backend strain, stale cache serves the base page and retries ESI requests.

• They use a TTL of 15 minutes with ESI blocks excluded from the cache. It helps ensure personalization works even when the main backend is under heavy load.

FAQs

1. How do I set the grace period in Magento 2?

You set the grace period in your Varnish VCL file using beresp.grace. Magento doesn't control Grace but generates the base VCL. Edit the generated VCL in vcl_backend_response to define grace durations per content type.

2. What happens when the Varnish grace period expires?

Once the grace period ends, Varnish will no longer serve stale content. If the backend is still down, the user may receive an error or a blank page. To avoid this, consider using a fallback page or custom error handling logic.

3. Can I use different grace periods for different pages?

You can customize grace periods by adding conditional logic in the VCL. Match URLs or backend responses to assign different beresp.grace values. It allows you to keep static pages cached for longer while limiting stale content.

4. How do I know if grace mode is working?

Use a varnishlog to check if Grace serves the responses. Look for indicators such as hit via grace or backend fetch delays. You can also add custom response headers in VCL to signal grace usage.

5. Can I disable grace mode for specific content?

Exclude specific URLs or headers using conditions in your VCL file. Set beresp.grace = 0s for those requests to skip stale content fallback. You can also control caching behavior using Magento’s cacheable.xml.

Summary

Varnish Grace Mode Magento 2 ensures high availability by serving expired cached content. The article explores the key points of the technique, including:

• Grace Mode serves stale content after the TTL expires to avoid delays or errors.

• ESI blocks in Magento help separate dynamic content from cached pages.

• Proper grace duration, cache invalidation, and monitoring are essential.

• Real brands like Everlast use Grace Mode to ensure reliability during outages.

Ensure uptime and smooth performance with Varnish Grace Mode. Choose managed Magento hosting for advanced caching and high availability.