Astralis or WNPower for rolling back WooCommerce cache settings

End a promotion without losing orders

A WordPress store in Asunción launches a forty-eight-hour promotion. Its Argentine provider works on a staging copy and changes the page cache to absorb catalog traffic. At closing, the reference price returns and the temporary rule must disappear. The store must prove that the public price is correct again, two open carts remain separate and test orders placed since the campaign began are still present.

This comparison concerns cache rollback. The change to reverse is a cache rule or setting identified before the campaign. The post-promotion price is already correct in WooCommerce. The team must bring public responses into agreement with that price while preserving subsequent writes. It keeps three independent proofs: source price, actually served price and order list. One screenshot after purging does not cover all three.

Astralis Básico offers an entry point in Paraguay with advertised OpenLiteSpeed and LiteSpeed Cache. WNPower Cloud Hosting Starter targets Argentina, advertises Apache with Nginx, a staging copy and a route-exclusion tool in cPanel. This pair compares two ways to remove a rule without changing site versions. Prices and limits are assessed in each offering’s market, in PYG and ARS.

Establish reference checks and a rollback time allowance

On protected staging populated with fictitious data, create a catalog item whose reference price is easily distinguished from its campaign price. Use a revision identifier and record expected prices for both states. Select a public page without personalized components to check caching. Record paths actually assigned to cart, checkout and account in WooCommerce. Translation or permalink changes can make generic exclusion lists incomplete.

In two independent browser profiles, put different items in carts A and B with different quantities. Keep a third anonymous profile without a cart for the public product page. On staging, create fictitious orders before and during the campaign and record their identifiers and statuses. Sensitive checks run on authorized test accounts and their own carts. No bank card or real address is used as a reference.

Set a deadline before testing. For example, allow ten minutes from the WooCommerce price change and finish with two consecutive passes, one minute apart, using the same public URL. The first observation proves the response changed. The second checks that a warmed cache still serves the expected price. The deadline is an acceptance criterion. Both hosts must be timed on a controlled installation if this window determines the contract.

Astralis Básico and restoring LiteSpeed settings

In Paraguay, Astralis advertises Enhance, OpenLiteSpeed, LiteSpeed Cache, SSL and daily backups. The table states 1 GB SSD, approximately 1 GB memory, a maximum of 1 vCPU, 500 IOPS and 90 simultaneous processes. The domain is billed separately. Storage must cover store media and data. Check space used by any copy and permission to create it. If the store approaches quota, staging needs separate space priced in the budget.

LiteSpeed Cache documents settings export and purging by URL, post identifier or category. Before the campaign, export settings from staging and record the plugin version and destination URL. At campaign end, preferably correct the identified setting if other settings changed meanwhile. A full import replaces every plugin setting. It suits cases where version and scope match and changes made during the campaign have been reviewed.

OpenLiteSpeed does not support ESI, which assembles page fragments governed by different caching policies. On this stack, treat responses containing a personalized cart or mini-cart as dynamic until separation is validated. LiteSpeed’s documented WooCommerce exclusions are a starting point. Check actual routes after restoring configuration. Daily backups help with other incidents. For this rollback, retain the active database.

Astralis advertises Spanish support. The site offers public channels and portal access. Enhance documents several interface languages, including French and Spanish in beta. Confirm the language in the account’s version and for a technical ticket. Consulted pages establish neither a campaign-end response time nor a ready-to-use isolated copy on Básico.

WNPower Starter and separating Nginx rules from purging

Cloud Hosting Starter advertises 2 cores, 2 GB dedicated memory, 40 GB SSD or NVMe, daily S3 backups and staging. The guide describes a WordPress copy. Staging must have an identifiable host and tested protection. When the copy is in a subdirectory, check that purging or rule changes target that path and the panel does not process the whole production store in the same operation. WNPower help describes JetBackup copies as best effort and sets no Starter-specific retention duration. Keep the cache-settings export with the acceptance record.

The HTTP Performance guide shows selecting a domain, subdomain or alias, then excluding a path in cPanel. Reactivar caché removes this exclusion. Documentation specifies that the field accepts a path without query parameters. Reserve this command for removing a temporary catalog exclusion. Private routes retain their protections. For variants depending on parameters or cookies, check the relevant layer’s rules and observed behavior on the staging domain.

WNPower also describes an Nginx purge in cPanel and a temporary-disable button that may reactivate after a few days. Purging removes stored responses. Exclusion governs future responses. Temporary server disabling has another scope. Record all three states so a technician taking over can distinguish them. If a WordPress plugin also caches HTML pages, handle its rule and responses before those of the Nginx proxy.

The page advertises Cloudflare integration and states both unmetered bandwidth on the card and 90 GB monthly traffic in the Starter FAQ. Ask which threshold applies and what headroom remains when cache is bypassed. Also check AWS instance and backup regions if they matter to the project. The consulted help center and labels are in Spanish, and customer support uses tickets. Confirm French and response time if the team depends on them.

Read amounts, taxes and renewal separately

Astralis displays 29,000 PYG per month or 290,000 PYG for twelve months on Básico. The annual cycle costs the equivalent of ten displayed monthly payments, a difference of 58,000 PYG compared with twelve monthly payments. Price the domain and any separate staging independently. The consulted table does not specify whether IVA is included. Ask for the total, tax breakdown and next-cycle price before comparing quotes.

Astralis terms provide for automatic renewal unless canceled before the next period. Record the exact expiry, request channel and date by which cancellation must be registered. Terms also allow price revisions. The recorded annual amount is a reference at verification time, with future pricing to confirm.

For WNPower, the Cloud Hosting Starter card displays a promotional 32,145 ARS excluding VAT per month. The FAQ displays 64,290 ARS excluding VAT per month at renewal and states Argentine IVA of 21%. Calculated corresponding monthly amounts are 38,895.45 ARS including VAT and 77,790.90 ARS including VAT. The card mentions two months saved with annual payment. Check the offered cycle, charged amount and combination conditions in the summary before calculating a twelve-month total.

WNPower terms state automatic renewal for the same duration as the selected cycle. Help explains where to view the next date and amount in the account. Under the terms, changing the period requires a ticket no later than five days before the next renewal. Cancellation requests go through the portal and must be processed before the next cycle under the displayed schedule. Determine the price actually billed and export useful data before service ends.

Purge from the source toward visitors

Before ending the campaign, inventory layers serving HTML on staging. Distinguish WordPress or LiteSpeed caching, the origin proxy, any CDN and the browser. A one-click Cloudflare offer does not prove the CDN stores HTML. An image-only cache does not retain the price displayed by the document. Keep a list of actually active layers and people able to purge them.

Validate the new price at origin and record rule states. Remove the campaign setting while checking that cart, checkout and account exclusions remain in force. Purge the layer nearest WordPress first, then the proxy, then the CDN if it actually caches HTML. If an external layer is purged while origin still serves the old product page, it can refill with the old price. Another external purge will then be required after correcting origin.

For currency, language or product variants, identify affected addresses and variants. The product URL is not always the only entry displaying its price. Catalog, category and other pages repeating the product may retain an old representation. Use a fresh control browser and revisit the same URL on the second pass. Compare response bodies and available headers. A cache signal at one layer does not describe other layers’ policies.

Sign off cart isolation and preservation of writes

Revisit carts A and B before and after purging. On account A, record its item and quantity. Repeat for B, then confirm a profile without a cart starts empty. Checkout and account pages must remain specific to the authorized profile. If the store uses WooCommerce blocks, also check dynamic responses feeding the cart and test quantity changes. For a personalized response, the expected record includes both correct content and no public reuse.

Check test orders using the same identifiers and statuses as before the intervention. Rolling back cache settings can alter visible responses without touching records. A general database restoration could instead erase newer orders. If an incident requires data recovery, handle it separately with a record of orders to preserve and a store-approved procedure.

Declare campaign exit validated if the reference price displays twice consecutively, each cart retains only its expected items, the anonymous account is empty and test orders remain present. Stop deployment if a personalized response is publicly reused, the price regresses between passes or an order is missing. Record the setting name, purged layer, time and author of each action. This helps attribute a result to an operation rather than expiration of a time to live.

Replay both purge orders offline

The supplied exercise simulates a catalog, two sessions and an order register. It uses two stores for public responses, origin and edge, and private data served directly by the mockup. It runs in Node.js without dependencies. Run the script after extraction. The first scenario removes the temporary setting, purges origin then edge and checks two consecutive public responses. A second purges edge before origin and introduces a visit between these operations. Open a terminal in the extracted folder, run “node exercice.cjs” and compare decisions with “attendu.json.”

This second journey shows how an outdated public page can reenter the outer cache while the register remains intact. After origin purging, an origin check returns the new price. Visitors passing through edge still see the previous price until a second purge there. Both carts remain separate throughout the test. The mockup never places private data in the two stores and does not reproduce LiteSpeed, Nginx or a real CDN’s operation.

Use the report to check criteria and explain failure. In a real trial, record date, time and where responses originate. If staging protection prevents all caching, validate isolation there and separately establish cache-warming evidence on an authorized test showcase containing only the intended fictitious data.

Choose according to the action the team can perform itself

Choose Astralis Básico for a lightweight catalog in Paraguay when Spanish support suffices, the copy fits the budget and the team understands LiteSpeed Cache settings. Its rollback route is targeted editing or controlled configuration import, followed by a page purge. If testing requires a second isolated environment, confirm its location, permissions and cost. Reject this scenario if the project assumes ESI fragment assembly with OpenLiteSpeed alone.

Choose WNPower Starter when the team wants advertised staging and a cPanel command removing a path-based Nginx exclusion. This choice is better documented for a team preferring to see exclusion status in the panel. If the rule concerns a parameterized request, the panel field cannot fully describe it. The decision then depends on which layer owns the rule and the access needed to remove and purge it.

If the rollback window is contractual, request a timed demonstration of the purge chain and permissions at each layer. Without evidence about external HTML caching, plan a campaign restricting page cache to the anonymous showcase and excluding personalized screens. Acceptance testing is ready when another person can find orders, reproduce checks and explain why both public passes display the same price.

Further reading

• Read the Astralis profile
• Read the WNPower profile
• Diagnose an inconsistent WooCommerce cart
• Download the cache rollback mockup (ZIP)