Modules

Web troubleshooting

My site won't open · my change isn't visible · I'm not on Google · my site is slow · the theme broke · my email stopped" — a diagnostic flow for six problems.

This guide closes the six most common Blesyum Web problems with measurable steps. Each section first asks "who produces the symptom": mistaking an infrastructure decision for a product fault delays looking in the right place.

Check the panel first. The Site health card on Overview names most of the problems below and says what to do. If the card is empty and does not say "could not measure", there really is no finding.

1) "My site won't open"

Ask three things in order; if the first answers, stop.

CheckMeaning
Sites → is the site Published?Draft/Maintenance/Archived deliberately keep visitors out. Suspended is a MANAGEMENT decision and needs support
Pages → is at least one page published?A site with no published page CANNOT be published; if it were, visitors would get a 404
What does the Domains screen say?Three distinct states with three different fixes: waiting for DNS · issuing certificate · live

⚠ The platform address (<slug>.web.<domain>) ALWAYS works. If it opens but your own domain does not, the problem is not the site but the address — look at the Domains screen.

⚠ If your domain says "not secure", DNS is verified but the certificate has not been issued yet. This takes minutes; reloading does not speed it up.

2) "My change isn't visible"

Every change you publish refreshes the server cache by itself (publish · unpublish · theme change · settings · bulk actions). In the normal flow there is nothing for you to do.

If the page still looks old:

  1. Browser cache — Ctrl/⌘ + Shift + R. This is by far the most common cause and has nothing to do with the server.
  2. Did you publish the right version? Publishing a page puts a specific version live: an edit left in draft NEVER reaches visitors. Confirm the row says Published and the latest version carries the live marker.
  3. Server cache — "Refresh server cache" on the Handover screen. This is not a delete: the key's generation advances and old entries expire on their own.

🔴 Do NOT make a habit of pressing that button after every change: the cache already refreshes itself and the button exists for a fault.

3) "I'm not on Google"

The product's own settings can switch off indexing, and this is the most commonly missed cause.

CheckWhere
Is search_indexing on?Settings (default ON)
Is the site published?If not, robots.txt is fully closed and no sitemap is produced
Is the site password protected?A password-protected site has both sitemap.xml and robots.txt closed — publishing its structure would apply that decision only halfway
Is the page noindex?Editor → Search engine (SEO) section
Does sitemap.xml contain the page?https://<your-address>/sitemap.xml

⚠ Indexing takes time. When the above are correct, patience is the answer; submitting the sitemap in Search Console helps. Publishing the same content on two addresses (www + apex) splits ranking — the product already pins one canonical address, so do not define a second one by hand.

⚠ If you changed a page path, the old address moves with a 301 and is not lost; do not delete it.

4) "My site is slow"

Visitor pages carry no JavaScript — so the classic cause (script weight) structurally does not exist here. Three causes remain:

  1. Large images. Uploading a 4000 px photo and drawing it at 320 px burns mobile data and LCP. The product generates ?w= derivatives and srcset automatically, but it never upscales beyond the source.
  2. Too many blocks. Hundreds of blocks on one page grow the HTML.
  3. Your own CDN. If you put a CDN in front, page HTML is served no-store (so a published change is never hidden) — the CDN does not cache the page, only speeds up assets.

Measure: node ops/web-performans-olcum.mjs <slug> (LCP · CLS · HTML size · request count). Exceeding the budget exits with code 1.

5) "The theme broke"

When a theme cannot render, the visitor does not see a blank screen: the product falls back to a safe skeleton and the page stays readable. But it is not silent.

  • Overview → Site health: "<path> failed to render N times in the last 7 days" plus the reason class (sablon_yok · sozdizimi · butce · veri). The four classes need different fixes, which is why they are reported separately.
  • Theme versions are kept: rolling back needs no re-upload, the package is already stored.
  • When uploading a new theme the validator reports problems at upload time — far cheaper than going live and hunting.

⚠ If your theme contains <script> or an image from an external CDN it fails silently in the browser: the visitor page security policy accepts neither. The validator reports this as an error at upload time.

6) 🔴 "My email stopped" (DNS / MX)

This is the riskiest side effect of domain operations and has nothing to do with the site itself.

When you delegate your domain's nameservers to Blesyum, its DNS records are answered from our zone. Any record not moved into that zone ceases to exist — and email works through the MX record.

Before delegating, write these down and recreate them exactly:

RecordPurposeIf skipped
MXIncoming mailNo mail arrives at all
TXT (SPF)Acceptance of mail you sendYour mail lands in spam
TXT/CNAME (DKIM)Signature verificationYour mail lands in spam
TXT (DMARC)PolicyYour domain becomes easier to spoof
CNAME/A (subdomains)webmail. · crm. etc.Those addresses stop opening

The DNS screen supports MX and its priority field is required (e.g. 10). Propagation after you add the record takes minutes to hours; during that window mail queues, it is not lost.

⚠ If mail is already down, the first move is NOT to revert the nameservers but to add the missing MX record: reverting also needs propagation time and lengthens the outage.

Still stuck?

The diagnostic sections on the Handover screen (link and asset audit · cache refresh) and the health card on Overview already produce what a support ticket needs. Additionally:

bash
blesyum-site health --json      # site status, machine readable
blesyum-site diagnose           # diagnostic bundle (carries NO secrets)

🔴 The diagnostic bundle carries no token, API key, database address or customer content: a tool that drops credentials into a support ticket creates a bigger problem than the one it solves.