For developers
The failing step and where to look, in the alert
When a store breaks at 2 a.m., you should not start from "the site is broken". StoreOnCall tells you which step failed, what changed just before, and which jobs, indexes or queues are unhealthy.
An alert written for you
Detail, not a vague page
Every alert is written twice. The owner gets plain words; you get this. Example alert.
[critical] Customers cannot complete checkout
Checkout test failed at "pay": payment step returned HTTP 500 (2 of 2 tries)
Likely cause: deployment at 02:08 (static content version changed)
Recent changes: Vendor_Payment 3.4.1 to 3.5.0
Check: var/log/exception.log, payment module credentials,
provider status page
Acknowledge: https://app.storeoncall.com/ack/...
Every rule, in the open
13 checks with exact thresholds
No black box. Each check says what it watches, when it alerts, and what you are told to look at.
Sales
Checkout test
- What it watches
- Every 15 minutes a test shopper adds a product to the cart, enters a test address and checks that shipping and payment options appear. It never places an order.
- Why it matters
- If the test shopper cannot reach payment, real customers probably cannot either.
- When it alerts
- A test fails twice in a row (we double-check before counting it). Critical, alerted immediately. If a firewall blocks the test, it shows as "could not test", never as a failure.
- Try to buy something on your store yourself, on your phone.
- Forward the alert to your developer: it says which step failed.
- Open the failing step (add to cart, checkout page, shipping or payment) and check var/log/exception.log.
- Check deployments, extension or setting changes listed under "Recent changes".
- Check the payment and shipping providers' status pages and API credentials.
Sales
Orders compared with normal
- What it watches
- Orders in the last 10 minutes to 8 hours, compared with what is normal for your store at that time of day and day of week, adjusted for how the last day has been trading.
- Why it matters
- A sudden drop in orders is often the first sign that something is broken, even when every page loads.
- When it alerts
- Orders are at most half of normal and the chance of such a drop being normal ups and downs is under 0.02%. Needs about 2 days of reports first, and windows expected to have fewer than 4 orders are never judged. Critical when orders stop completely, high otherwise.
- Check the checkout test result on this page.
- Try to place a test order yourself.
- Check whether a campaign ended or ads were paused: those lower orders without anything being broken.
- Check checkout, payment methods and recent deployments.
- Compare traffic in your analytics: if visitors are normal but orders are not, the problem is on the site.
Sales
Payment methods
- What it watches
- For each payment method, how many orders in the last 3 hours ended up stuck waiting for payment or cancelled. Orders under 30 minutes old are left out, because customers may still be paying.
- Why it matters
- When a payment method breaks, customers who choose it cannot pay, and the orders sit unpaid.
- When it alerts
- A method has at least 3 orders in the last 3 hours and half or more of them are stuck or cancelled. High, alerted immediately.
- Check your payment provider's dashboard and status page.
- Consider switching that payment method off until it is fixed.
- Check the payment module's logs and API credentials.
- Check webhooks or IPN callbacks from the provider reach the store.
Background jobs
Background jobs (cron) running
- What it watches
- The plugin reports through Magento's background jobs (cron) every 5 minutes, so a missing report means cron has stopped or the server cannot reach us.
- Why it matters
- Without cron, prices, stock, search, emails and indexes stop updating.
- When it alerts
- No report for 20 minutes. High, alerted immediately.
- Forward the alert to your developer or host.
- Check the Magento cron line in crontab -l and var/log/cron.log.
- Run bin/magento storeoncall:status on the server.
- Check the server can make outbound HTTPS calls.
Background jobs
Failing background jobs
- What it watches
- Background jobs that ended in an error in the last hour.
- Why it matters
- Repeated failures usually mean emails, indexes or imports are not happening.
- When it alerts
- 3 or more failed jobs in the last hour, lasting 15 minutes. Medium.
- Forward to your developer; no action needed from you.
- Check cron_schedule for status=error and the messages column.
- The alert lists the job codes.
Background jobs
Skipped background jobs
- What it watches
- Background jobs Magento skipped in the last hour, usually because the server is overloaded or cron runs too rarely.
- Why it matters
- Skipped jobs mean updates happen late or not at all.
- When it alerts
- 10 or more skipped in the last hour, lasting 30 minutes. Medium.
- Forward to your developer.
- Check cron frequency (every minute), server load and long-running jobs.
Background jobs
Stuck background jobs
- What it watches
- Jobs that have been "running" for over an hour.
- Why it matters
- A stuck job can block the same job from running again, for example price or stock updates.
- When it alerts
- Any job running for over an hour, lasting 30 minutes. Medium.
- Forward to your developer.
- Find the job in cron_schedule (status=running) and check whether its process is still alive.
- Clear stale running entries once the cause is fixed.
Prices, stock and search
Indexes up to date
- What it watches
- Magento's indexes, which keep prices, stock, categories and search results correct on the storefront.
- Why it matters
- An out-of-date index can show wrong prices, sell out-of-stock products or hide products from search.
- When it alerts
- Any index marked invalid for 30 minutes. High when it is prices, stock, search, categories or catalog rules; medium otherwise.
- Forward to your developer.
- Spot-check a few product prices and stock on the storefront.
- Run bin/magento indexer:status, then bin/magento indexer:reindex for the listed indexes.
- Check why they were invalidated: a failed reindex, cron, or an admin save.
Prices, stock and search
Index updates waiting
- What it watches
- For indexes set to "update by schedule", how many changes are waiting to be applied.
- Why it matters
- A growing backlog means storefront prices and stock fall behind what is in the admin.
- When it alerts
- 5,000 or more waiting, or 500 or more and growing by half within 30 minutes, lasting 30 minutes. Medium.
- Forward to your developer.
- Check the indexer_update_all_views and indexer_reindex_all_invalid cron jobs and the mview changelog tables.
Background jobs
Message queues
- What it watches
- Magento's message queues (for example bulk actions, inventory and exports).
- Why it matters
- Stuck queues delay stock updates, bulk changes and exports.
- When it alerts
- A queue with 1,000 or more waiting and still growing, or new errors, lasting 30 minutes. Medium.
- Forward to your developer.
- Check consumers (bin/magento queue:consumers:list, cron consumers_runner) and RabbitMQ if used.
Background jobs
Order emails
- What it watches
- Order confirmation emails waiting to be sent (only when Magento sends emails in the background).
- Why it matters
- Customers who do not get a confirmation contact support or order again.
- When it alerts
- 5 or more order emails not sent, lasting 30 minutes. Medium.
- Forward to your developer.
- Check the sales_send_order_emails cron job and the mail server or SMTP extension logs.
Speed
Page cache
- What it watches
- Whether Magento's full page cache is switched on.
- Why it matters
- With the page cache off, pages load much more slowly and the server can struggle at busy times.
- When it alerts
- Page cache switched off for 10 minutes. High.
- Forward to your developer.
- Run bin/magento cache:enable full_page and check why it was switched off.
Speed
Caches needing a refresh
- What it watches
- Cache types Magento has marked as needing a refresh after changes.
- Why it matters
- Usually harmless; if it lasts, storefront changes may not show.
- When it alerts
- For 60 minutes. Low: shown here and sent to developers only, never to owners.
- Nothing to do.
- Refresh the listed cache types (bin/magento cache:clean <types>).
Sales
Checkout test
- What it watches
- Every 15 minutes a test shopper adds a product to the cart, enters a test address and checks that shipping and payment options appear. It never places an order.
- Why it matters
- If the test shopper cannot reach payment, real customers probably cannot either.
- When it alerts
- A test fails twice in a row (we double-check before counting it). Critical, alerted immediately. If a firewall blocks the test, it shows as "could not test", never as a failure.
- Try to buy something on your store yourself, on your phone.
- Forward the alert to your developer: it says which step failed.
- Open the failing step (add to cart, checkout page, shipping or payment) and check var/log/exception.log.
- Check deployments, extension or setting changes listed under "Recent changes".
- Check the payment and shipping providers' status pages and API credentials.
Sales
Orders compared with normal
- What it watches
- Orders in the last 10 minutes to 8 hours, compared with what is normal for your store at that time of day and day of week, adjusted for how the last day has been trading.
- Why it matters
- A sudden drop in orders is often the first sign that something is broken, even when every page loads.
- When it alerts
- Orders are at most half of normal and the chance of such a drop being normal ups and downs is under 0.02%. Needs about 2 days of reports first, and windows expected to have fewer than 4 orders are never judged. Critical when orders stop completely, high otherwise.
- Check the checkout test result on this page.
- Try to place a test order yourself.
- Check whether a campaign ended or ads were paused: those lower orders without anything being broken.
- Check checkout, payment methods and recent deployments.
- Compare traffic in your analytics: if visitors are normal but orders are not, the problem is on the site.
Sales
Payment methods
- What it watches
- For each payment method, how many orders in the last 3 hours ended up stuck waiting for payment or cancelled. Orders under 30 minutes old are left out, because customers may still be paying.
- Why it matters
- When a payment method breaks, customers who choose it cannot pay, and the orders sit unpaid.
- When it alerts
- A method has at least 3 orders in the last 3 hours and half or more of them are stuck or cancelled. High, alerted immediately.
- Check your payment provider's dashboard and status page.
- Consider switching that payment method off until it is fixed.
- Check the payment module's logs and API credentials.
- Check webhooks or IPN callbacks from the provider reach the store.
Connection
Store data can be read
- What it watches
- Every 5 minutes StoreOnCall reads the store's recent orders through the read-only BigCommerce app.
- Why it matters
- If we cannot read the orders, we cannot tell whether customers are buying.
- When it alerts
- No successful read for 20 minutes. High, alerted immediately. Removing the app stops reading without an alert.
- Check the StoreOnCall app is still installed: Apps > My Apps in BigCommerce.
- Check the BigCommerce status page.
- Reinstall the app if its access was revoked.
Install
Ten minutes, read-only, no customer data
A standard Magento module. It adds no tables, pages or admin users, rides on Magento's own cron, and sends counts and statuses every 5 minutes. Preview the exact report before it is sent.
composer require storeoncall/module-agent
bin/magento module:enable StoreOnCall_Agent
bin/magento setup:upgrade
bin/magento cache:flush
bin/magento storeoncall:preview
bin/magento storeoncall:connect <store key>
Fits your tools
Alerts where you already work
How signals travel
From the store to your phone
Your store. Magento or Adobe Commerce, on your server or Adobe Cloud. Nothing is changed, and no customer data leaves it. Click any box.
Test shopper. Every 15 minutes it adds a product to the cart, enters a test address and checks shipping and payment options appear. It never pays, and deletes its cart. A failure is checked twice.
Read-only plugin. Every 5 minutes, through Magento's own cron, it sends counts and statuses: orders, payments, cron, indexes, queues, order emails and caches.
StoreOnCall. Every minute it compares what came in with what is normal for your store at that hour and day, learned from your last 28 days. Only problems that are confirmed and last become incidents.
The owner gets plain words and the sales at risk per hour, on email, WhatsApp, Slack or Teams. One tap to acknowledge, or it escalates.
The developer gets the failing step, recent changes and where to look, on Slack, PagerDuty, Teams, a webhook or email.
Dashboard and reports. Every store on one screen, incident timelines, and a monthly reliability report to share.
Click any box
Blue is the test shopper, green the plugin's reports, red the alerts. Nothing flows back into the store.
Try it on a staging store
Free scan first, then a trial with the plugin on staging.