Email alerts

Guardioz sorts its checks into groups, and one group covers one kind of problem on one server, for example Autovacuum or Invalid indexes. Guardioz sends one kind of email about your servers: a message when one of these groups appears, or becomes more severe, at a severity level that you chose to receive and that your plan includes. The three levels are critical, warning and suggestion - see Read your findings. When you change nothing, you receive email about critical groups only.

The plan of your organization decides which levels can send email. The Free plan sends email about critical groups only. The Team and Scale plans send email at every level: critical, warning and suggestion. This applies now, during the beta.

This page says how you choose the levels, when the message goes out, who receives it, what is in it, and what it never reports.

Choose the levels

Sign in at https://app.guardioz.com/ and click your email address in the top navigation. The Account Settings page opens.
In the section Notifications, select the levels you want email about: Critical, Warning and Suggestion. Until you change it, an account has Critical selected and the other two not selected.
Click Save. The page confirms with Notification settings updated.

The choice belongs to your account and applies to every server of your organization. There is no choice per server. With no level selected, Guardioz sends you no email about findings.

On the Free plan you can select Critical only. Warning and Suggestion are shown greyed out, with the label Team plan, and a sentence under the levels names the plan of your organization and the address to write to about the Team plan: contact@guardioz.com. No page in Guardioz changes the plan: Guardioz changes the plan of an organization, so write to that address about it. When your account selected Warning or Suggestion before, Guardioz keeps that choice but sends no email at those levels while the organization is on the Free plan. The boxes then show as not selected. When the organization moves to the Team or Scale plan, the kept choice applies again, and the next analysis reports the groups that already stand at those levels.

When you select a level at which groups are already open, the next analysis of each server that covers those groups reports them to you once, provided Guardioz has not reported them at that level before. With one snapshot per hour, that is within about an hour for most groups. Invalid indexes and Duplicate indexes are analysed only when the catalog of tables and indexes arrives. After the first 24 hours that is about once a day, so for these two groups it can take about a day, or less when you run a collection by hand.

When Guardioz sends the message

Guardioz analyses every snapshot that reaches it. After all the applicable checks have finished, it compares what this analysis found with what it has already reported for that server. Guardioz then emails you about a group when both of these are true:

the severity of the group in this analysis is higher than the level Guardioz last reported for it on this server - a group that was not reported before counts as higher;
that severity is one of the levels you selected.

One message covers one server and one analysis. When several groups qualify in the same analysis, all of them stand in the same message, the most severe first.

The first snapshot is sent during the installation, so a first message can arrive minutes after you install the agent.

While a group stays at the level it was reported at, Guardioz sends no second message. Guardioz remembers, for each server and each group, the level it last reported. A group that becomes more severe, for example from warning to critical, is reported again at the new level. That memory is reset when a later analysis finds the group clear, or finds it at a lower severity. After a reset the group counts as not reported: when a group falls from critical to warning and you selected Warning, the analysis after the one that saw the fall reports it at warning. When a group that was solved comes back, a new message goes out.

Guardioz keeps that memory per server and group, not per recipient. A group counts as reported once the mail provider accepted a message that carried it.

Some checks do not apply to every server. A check that reaches no verdict changes nothing: Guardioz sends no message for it, and it does not reset what Guardioz has already reported. When a check cannot finish during an analysis, Guardioz sends no message for that snapshot. Every finding stands on the Findings page, so look there when you expect a message that did not arrive: see Read your findings.

The levels each group can reach

Every group can send a message, at the levels it can reach:

Group	Levels it can reach
Reliability settings	Critical
Autovacuum	Critical
Monitoring visibility	Warning
Performance tuning	Warning, Suggestion
Invalid indexes	Critical, Warning
Duplicate indexes	Critical, Warning
Tables behind on cleanup	Warning
Replication slots	Critical, Warning, Suggestion
Transaction log archiving	Critical, Warning, Suggestion
Open transactions and cleanup	Warning, Suggestion

With only Critical selected, and on the Free plan, the four groups that do not reach the critical level send you nothing. For what each group looks at, see What Guardioz checks.

Who receives the message

Guardioz sends it to every user of your organization whose email address is confirmed and whose account is not disabled. Each recipient gets one message with the groups at the levels that recipient selected, and no message when none of the groups is at one of those levels. When no address qualifies, Guardioz sends nothing.

Today an organization holds one account, so that address is your own. Guardioz has no way to invite another person into your organization, so a second recipient cannot be added. See Manage your account.

What the message contains

The message comes from Guardioz at noreply@guardioz.com. It carries no reply address, so write to contact@guardioz.com when you need an answer from us.

The subject counts the groups and names the server:

2 critical issue groups on SERVER_NAME - Guardioz

SERVER_NAME is the name of the server in Guardioz. When the message holds one group, the subject reads 1 critical issue group. A message that holds groups at more than one level counts each level, the most severe first:

1 critical and 2 warning issue groups on SERVER_NAME - Guardioz

The body holds:

the same count as a heading, with the server name;
the sentence This report comes from the analysis of the snapshot collected at, with the time of that collection. This time is always in UTC, not in the time zone of your account;
one row per group: the group title, and how many objects the group affects, for example 3 affected objects. An object is the thing a check looked at: a setting, an index, a table, a replication slot, the archiver, or a database for Open transactions and cleanup. That number counts objects inside one group, while the number in the subject counts groups. In a message that holds more than one level, each row also names its level;
for six groups, an explanation under the row, for each object that stands at the level of its group. For Reliability settings, Autovacuum, Monitoring visibility and Performance tuning it is the same text the setting’s page shows under Why this value?. For Transaction log archiving it is the sentence the archiver page shows under What we found. For Replication slots it is the name of the slot and the two sentences its page opens with, under What we found and Why it matters; the message does not carry the slot’s database, the statement that drops it, or the step that says what to check first - those stay on the slot’s page. The explanation is the same at every level: a warning or a suggestion row carries it the same way a critical row does. The rows of the other four groups carry the title and the count only, at every level;
when the message explains part of a group, a sentence that says so, for example This message explains 3 of the 12 affected objects.. A message explains at most 5 objects of one group, and it leaves out an object whose level differs from the level of its group - for example a suggestion inside a Performance tuning group that stands at warning. To see every object of a group, open the group from the Findings page;
a button See what needs attention, which opens the Findings page for that server. For a message with one level, the list is filtered to that level. For a message with more than one level, it shows every open issue, the most severe first. You must be signed in to open it;
the closing line Guardioz keeps watching this server and reports what changes. The message reports what this page describes: a group that appears or becomes more severe, at the levels you selected;
a footer with the company details of the provider, the address contact@guardioz.com for questions, and links to the Terms of Service and the Privacy Policy.

The message holds no unsubscribe link and no link to your settings. To stop it, clear the levels in Account Settings: see Stop the messages.

What Guardioz never emails about

A level you did not select. Until you change the levels, you receive email about critical groups only.
A level your plan does not include. On the Free plan that is warning and suggestion, also when your account selected them before.
The moment a problem is solved or a group becomes less severe. Guardioz records the change and sends nothing in that analysis. A group that became less severe then counts as not reported, so when you selected its new level, the next analysis that finds it there reports it at that level.
An agent that stopped sending, a server whose snapshots no longer arrive, or a snapshot that Guardioz refused. No email tells you that data stopped arriving, so read Last seen on the Servers page, or the server page, yourself: see When snapshots stop arriving.
A daily or weekly summary. Guardioz has none.

This message is the only email Guardioz sends about findings. The Findings page carries the findings at every severity, including the ones that send you no message. See Read your findings.

Delivery is not guaranteed

An email can fail to arrive, and it can arrive more than once. Guardioz knows only that the mail provider accepted the message. It does not follow what happens after that, so a message that your spam filter catches still counts as accepted. When you expect a message and none arrived, look in the spam folder of your mailbox, and allow noreply@guardioz.com in your mail filter.

When the mail provider accepts the message for no recipient, Guardioz records nothing, and the next analysis that still finds the group at that level tries again. With one snapshot per hour, that is about an hour later for most groups, and up to about a day later for the two index groups, which are analysed when the catalog arrives.

Read email as a second signal, not as the only one. Everything a message reports is on the Findings page.

Stop the messages

To stop every email about findings, clear all three levels in the section Notifications of Account Settings and click Save. See Choose the levels. There is no way to silence one server or one group, and no way to send the message to Slack or to a webhook target. A weekly summary email, and messages to Slack, Microsoft Teams and webhooks, are coming in autumn 2026.

These actions also stop the messages:

Solve the problem. When a group is clear, Guardioz sends nothing about it until it comes back. See Apply a recommendation.
Stop the agent on that server. Guardioz analyses a snapshot when one arrives, so a server that sends nothing produces no analysis and no message. That server is then no longer monitored. See Manage the agent service.
Delete the server in Guardioz. See Manage your servers.
Delete your account. This removes the only recipient together with your organization, its servers and what Guardioz collected for them, and it cannot be undone. See Manage your account.

Next steps

Read your findingsFilters, severities and states on the Findings page.

Apply a recommendationWhere to run a recommended change and when the finding changes afterwards.

What Guardioz checksOne entry per check group, with the data it looks at.