Short answer: Partly, and the ceiling on that check is worth knowing before you pay. Three things are public, free and readable without an account: whether the provider blocks outbound mail ports by default and exactly what it takes to lift the block; who is allowed to set the PTR record for the address you will be given and whether you can change it; and whether addresses in the provider's range appear on public DNS blocklists. One thing is not checkable: the reputation of your own address, because no address is allocated until the machine is provisioned. So the pre-purchase check tells you the rules and the neighbourhood, never the verdict on the single IP you will receive. Those three answers live in support documentation, DNS standards and public lookup tools, and they disagree with each other considerably more than a plan table suggests.
| Part of outbound identity | Verifiable before paying | Where the answer is published |
|---|---|---|
| Mail ports | The default outbound state, and the precise request that changes it | Four providers' support pages [S1][S2][S3][S4] |
| Reverse DNS | Who creates the PTR, whether you can edit it, and the standard a usable entry has to satisfy | Provider DNS documentation and two RFCs [S5][S6][S7][S8] |
| Address-range reputation | Whether public lists contain the range, and what a listing actually asserts | Blocklist policies and multi-list lookup tools [S9][S10][S11][S12][S13] |
Four large publishers put their outbound rules in writing. The rules are not interchangeable.
| Publisher | Default outbound state | What changes it | Source |
|---|---|---|---|
| DigitalOcean | Ports 25, 465 and 587 are blocked on Droplets by default, including traffic that passes through a Reserved IP | That page publishes no unblocking route; it recommends sending through a third-party email service instead | [S1] |
| Google Cloud | Destination TCP port 25 is blocked when the destination is outside the VPC, although some projects are exempt and traffic within a VPC, a peered VPC, or a network attached by Cloud VPN or Cloud Interconnect is allowed | Relaying through Google Workspace on port 465 or 587, or a firewall rule that permits a custom port such as 2525 | [S2] |
| AWS | Outbound port 25 is blocked by default on EC2 instances and Lambda functions attached to elastic network interfaces | A submitted request to remove the email sending limitation; the same form can carry an rDNS record for an Elastic IP | [S3] |
| Hetzner | Ports 25 and 465 are blocked for new customers | After one month and a paid first invoice a limit request may be filed; port 587 is not blocked and requires no request | [S4] |
Two details here are easy to miss because they break the usual "just use submission" advice. The first is that one provider blocks all three commonly named mail ports, including 587, while another explicitly leaves 587 open and says so in the same paragraph as its restriction. [S1][S4] A buyer who assumes 587 is universally available will discover otherwise only after provisioning. The second is that a block can be conditional on account history rather than on configuration: one provider ties the lifting of a port restriction to a billing event. [S4] That is a time delay, not a setting.
RFC 5321 defines the SMTP transfer service that port 25 carries between mail transfer agents. [S14] It does not oblige any network to carry that traffic. Nothing in the standard prevents a provider from filtering it, which is why the only reliable source for the default is each provider's own published policy rather than a general description of how SMTP works.
The divergence matters for a cheap tier in a specific way. A low monthly price buys compute, memory and bandwidth; it does not buy an outbound mail path. Where the port is filtered, the documented alternative is submission to someone else's service or a relay, which introduces a second bill or a second account that no plan table lists. The port policy is therefore a cost question disguised as a firewall question.
A PTR record maps an address back to a name. Three providers give three different answers to the question of who controls it.
DigitalOcean says it creates PTR records automatically from the name given to a Droplet in the control panel, and gives a worked contrast: a fully qualified domain name as the Droplet name produces a PTR record, whereas a hostname-style label does not. It also states that on a Droplet with IPv6 enabled, only the first of the 16 available IPv6 addresses gets a PTR record, and that PTR records cannot be created manually through that control panel. [S5] The practical consequence is that naming a machine at provisioning time is a DNS decision, and on IPv6 the remaining addresses in the assignment have no reverse entry from that mechanism.
Hetzner's cloud documentation puts the control in the customer's hands: the rDNS entry is edited next to the IPs under the server's Networking section in its console. The same page defines a valid entry as one that resolves in both directions, so that the domain assigned to an address resolves back to that same address, and names this Forward-confirmed reverse DNS (FCrDNS). [S6]
AWS ties the two earlier threads together. Its port restriction article describes an optional Reverse DNS record field in the request form used to lift the port 25 limitation, and notes that an rDNS request cannot be submitted from GovCloud (US) Regions. [S3] On that platform the port question and the PTR question are handled in one workflow, which is not true at either of the other two.
RFC 1912 explains why the direction of the mapping is not cosmetic. It says every Internet-reachable host should have a name, that a matching PTR record should exist in the in-addr.arpa domain for every IP address, that a multi-homed host needs a PTR for each address rather than only the first, and that PTR records must point back to a valid A record rather than to a CNAME alias. It warns that failing to keep PTR and A records matched can cause loss of Internet services similar to not being registered in DNS at all. [S7] Read beside the IPv6 sentence above, that is a concrete gap rather than a theoretical one: an address with no reverse entry is exactly the case the RFC describes.
RFC 2317 supplies the part most buyers never see. It describes classless in-addr.arpa delegation, a method for delegating reverse authority for address spaces smaller than 256 addresses, because a single zone can only be delegated once. [S8] Reverse authority for a small cheap-tier allocation can therefore sit with a different party than the one selling the virtual machine. "Set my PTR" is not automatically a request the machine vendor can fulfil on its own.
The pre-purchase version of this section is a short list of questions to put to a provider's documentation: is a PTR created automatically or by request, can I edit it, does it cover every address I will receive including IPv6, and who holds the reverse zone. All four answers are published for the providers above. [S3][S5][S6][S8]
A DNS blocklist is a DNS zone queried with a reversed address. RFC 5782 documents the structure and the query protocol for DNS blacklists and whitelists, and states plainly that it describes structure, operation and use but does not describe or recommend policies for adding or removing addresses. [S9] That boundary is the single most useful fact for a buyer: the wire format is standardised, the criteria for getting on or off a list are not.
Spamhaus makes the distinction explicit. Its Policy Blocklist lists address space that should not be sending mail direct-to-MX, and can include both dynamic and static addresses depending on the block owner's policy; it is a statement about how an address is expected to be used, not a record of spam observed from it. [S10] Its separate Spamhaus Blocklist addresses IPs that appear to be under the control of, used by, or made available to spammers and abusers. [S11] A lookup result therefore needs the list name attached to it before it means anything. "Listed" on a policy list and "listed" for observed abuse are different claims.
The removal criteria on the policy list are unusually concrete and worth reading before ordering, because they describe what a clean mail server is expected to look like. The FAQ says an IP should only be removed if a mail server will run on it, appropriate forward and reverse DNS are set up, and outbound port 25 is closed for use by any device other than that mail server. It adds that single-IP exclusions are for static addresses assigned to the person performing the removal, that end-user exclusions expire after one year, and that removing many IPs is a task for the ISP holding the assignment. [S12] Those are the same three threads as this article - ports, PTR, and who controls the range - restated as a listing policy.
Free public tools exist for the lookup itself. MXToolBox says its blacklist check tests a mail server IP against over 100 DNS-based email blacklists. [S13] DNSBL.info describes checking an IP against more than 100 DNS-based blacklists on one page. [S15] MultiRBL describes itself as a multiple DNSBL lookup and FCrDNS check tool, and warns that neutral entries are informational only and do not mean mail is being blocked. [S16] That last caveat is worth keeping: a multi-list page showing a coloured row is not by itself evidence of delivery failure.
Check one, before any account: read the port policy. Search the provider's own documentation for outbound SMTP and port 25 rather than a review of the plan. Write down the default, the exact ports covered, and whether the unblocking route is a form, a support ticket, a billing milestone, or nothing published at all. [S1][S2][S3][S4] If the application needs direct outbound SMTP and the policy offers no route, the plan is unsuitable regardless of its price.
Check two, before any account: read who owns reverse DNS. Find the PTR or rDNS documentation and note whether the entry is generated from a name you choose, edited in a console, or attached through a request form. [S3][S5][S6] Then check whether that control covers every address in the assignment. [S5][S7] If the answer is not published, the honest pre-purchase conclusion is that you cannot verify it, not that it is unavailable.
Check three, when you have an address: run the lookups. Once an IP exists, a multi-list check shows whether it appears on public lists, and an FCrDNS check shows whether forward and reverse resolution agree. [S13][S15][S16] Before an IP exists, the same tools can only be pointed at addresses you already know, such as one quoted in a provider's own documentation or one you hold elsewhere. That yields information about a range, not a verdict about your future machine. Lookup counts from these tools are counts of lists queried, not measurements of deliverability. [S13][S15]
The ceiling deserves to be stated plainly. Reputation attaches to an allocation you do not yet have, on a range shared with customers you cannot see, under policies maintained by third parties who do not publish a common removal standard. [S8][S9][S10] No amount of pre-purchase checking converts that into certainty. What the checks do buy is the ability to rule out a plan whose published rules already make the workload impossible.
Once you have checked outbound mail identity checks, the promo renewal cost page traces the second-year price back to published billing terms.
While this page deals with outbound mail identity checks, the real cheap VPS complaints page reads the complaints that repeat across cheap-plan reviews.
The figures here cover outbound mail identity checks; for the part that follows, the delivered VPS spec evidence page sets out what counts as proof when a delivered machine does not match its plan page.
These are self-reported comments from a public discussion index, retrieved on 2026-11-11, each linked with its author and date. They are anecdotes, not measurements, and none comes from the three sites in the comparison set.
One operator describes the two AWS workflows as a single task: "I've run mail servers on EC2 before. You'll need to raise a request to remove the SMTP 25 restriction (both inbound and outbound), and also apply for a DNS reverse ptr update for that IP for it to work." [HN1] A second identifies the missing PTR as the actual cause of rejected mail: "The main problem I had with getting emails accepted was my lack of reverse DNS PTR records on my domain." [HN2] These two reports line up with the documentation's structure: the port and the reverse entry are separate switches that have to be set together. [S3][S7]
A third report is about the range rather than the individual machine: "blacklists can encompass the entire /24 of IPs that your single server resides in. You have no control over the abuse of other VM/VPS tenants with adjacent IPs in the same block and may never successfully get it taken off some RBLs." [HN3] A more recent comment frames the same risk as indirect and use-case dependent, saying an IP range "gets blacklisted (because some Hetzner clients abuse it for Sybil attacks or spam)" while adding that this is unlikely to affect an individual VPS's performance. [HN4] Those two do not contradict each other: one is about mail acceptance, the other about machine performance.
A 2015 comment describes the port default as something discovered rather than disclosed: "One thing to watch out for is that outgoing TCP connections to port 25 are blocked until you ask them to open it up." [HN5] It refers to a different provider from the four documented above and is quoted only as an example of the discovery pattern; it is not used as evidence of that provider's current policy.
The consensus across these reports is that outbound mail fails at the port, the reverse entry, or the neighbouring addresses - rarely at the application. The divergence is about how recoverable each failure is: the first two are usually a form or a console edit, while the third depends on a list maintainer and on other tenants' behaviour. [HN1][HN2][HN3]
A rule learned at one provider should not be carried to the next. Port 587 is open without a request at one publisher and blocked by default at another. [S1][S4] A PTR is generated from a machine name at one, edited in a console at a second, and attached through a port-unblock form at a third. [S3][S5][S6] A policy listing is a statement about expected use, while an abuse listing is a statement about observed behaviour, and a single lookup tool can display both kinds of row. [S10][S11][S16]
That variety is why "does it support email?" is the wrong question to ask of a cheap plan. The answer depends on which of the three parts a given workload needs. A site that sends password resets through an external relay needs a submission path and nothing else. A host that must deliver directly to other mail transfer agents needs the port, a matching reverse entry, and a range whose listing policy it can live with. Those are three different purchases.
Several documentation paths tried for this article returned HTTP 404 or 403, and one large documentation page returned a shell with no rDNS body in the retrieved HTML; none of them is quoted or used as a source. The OVHcloud reverse-DNS documentation was not retrievable in a usable form, and the Oracle page retrieved was an Email Delivery overview that publishes no default outbound port state, so no port claim is attributed to either. [S17][S18] The three comparison sites' editorial pages were checked for coverage of this sub-question, not mined for wording; no sentence from them appears below.
No deliverability test, no SMTP connection attempt, no message was sent, and no IP belonging to the site publishing this article was looked up. No address was allocated, so no statement about the reputation of a specific cheap-tier IP can be made. No figures from analytics tools, no traffic or search-volume estimates, and no ranking judgement appear in this article; the two data sources reserved for the 28-day report were not read.
The useful pre-purchase question is not "is this IP clean?" but "which of these three things can I still change after I have paid?" On the pages read, port policy and PTR control are published and sometimes adjustable, while the range you land in is assigned rather than chosen. That is the part of a cheap machine's identity no plan table shows, and the part worth checking while the tab is still free.
No. An address is assigned at provisioning, so before that moment there is no specific IP to query. Public lookup tools can be pointed at addresses you already know, which tells you something about a range, but it cannot produce a verdict on the address you will receive. [S13][S15]
It depends on the provider. One creates it automatically from the name given to the machine and does not allow manual creation through its control panel; one lets the customer edit it in the console; one attaches it through the form used to lift its port restriction. A valid entry resolves in both directions, and the reverse zone for a small allocation may be delegated to a party other than the machine vendor. [S3][S5][S6][S8]
No. One provider blocks 25, 465 and 587 by default on its virtual machines, while another states that 587 is not blocked and needs no request. Check the provider's own policy before assuming a submission port is available. [S1][S4]
All pages read on 2026-11-11. Every bracketed number in the text maps to one line below.
The three sites in the comparison set were examined for topic coverage only; their phrasing is not carried into this text, and no sentence here comes from a machine-generated transcript. Community comments are quoted verbatim with attribution and are self-reported. Derived statements are labelled as such. The two analytics data sources reserved for the 28-day report were not read for this article.