This completes the data/functionality migration for multiple DNS writers.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=163835077
Note that even though the nomulus command line tool now supports multiple
DNS writers for all subcommands, this still won't work quite yet because
the DNS task queue format migration from [] is still in progress.
After next week's push that migration will be complete and we can remove
the final restriction against only having one DNS writer per TLD.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=162490399
After this point all data is migrated to use the new canonical
plural version, and subsequent code changes can be made that use
multiple writers.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=161673486
This is the first step in a multi-step data migration to allow multiple
DNS writers per TLD. The overall process looks like this:
1. Add a plural DNS writers field with backfill (this commit).
2. Deploy it.
3. Run the ResaveEnvironmentEntitiesCommand to populate this new field
on all entities.
4. Update the code to use the new field everywhere.
5. Deploy it.
6. Delete the now-unreferenced, old deprecated singular value field.
This process is rollback-safe.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=161253436
Memcache is already off but now it's not in the code anymore.
This includes removing domain creation failfast, since that is actually
slower now than just running the flow - all you gain is a non-transactional
read over a transactional read, but the cost is that you always pay that
read, which is going to drive up latency.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=158183506
We ran into a bunch of prober deployment issues this past week when
attempting to spin up a new cluster because the newly created prober
TLDs had null values for the dnsWriter field. Given that VoidDnsWriter
exists, we can require that dnsWriter always be set, and have people
use that if DNS publishing is not required.
Also cleans up a bunch of related inconsistent exception messages and
tests not verifying said exception messages properly.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=154325830
TESTED=The test fails if you change line 134 in Ofy to not use memcache
and use the unchanged original Registry.get() code. This is the
expected behavior.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=154226534
This is better than calling assertTldExists() inside a for loop because you can throw a single exception reporting all bad TLDs at once rather than only getting as far as the first failure. And then it's also a one-liner instead of 3 lines.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=152412876
This is a follow-up to Lai's refactoring of the get reservation types
code to return a set rather than a single type. Since we're always
returning a set now, the more natural way to represent a label that is
not reserved is to return an empty set rather than a set containing
UNRESERVED.
Also fixes some minor style issues I ran across regarding static
importing and test method naming that I ran across (no logic
implications).
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=151132116
During domain create/applicationcreate/allocate, domains that are on the reserved list(s) with nameserver restricted reservation type must set nameservers that are part of the allowed nameservers for that domain in the reserved list(s) applied to that TLD.
Additionally a boolean is added to Registry to indicate if a TLD is restricting domain create. If it is, only domains that are nameserver restricted can be registered.
For consistency with a similar feature that validates a TLD-wide nameserver whitelist, the per-domain nameserver validation is performed even when the operation is in super-user mode. Similarly, if a domain is nameserver restricted, nameservers must be supplied (i. e. the nameservers set cannot be empty) when registering the domain.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=150641269
A new field (allowedNameservers) is added to ReservedListEntry that stores the allow nameservers for the label. The field itself is a comma separated string, but the actual lines within a reserved list file (from which the field is parsed) uses colon to separate nameservers, to avoid conflicting with the commas used as primary separators in a CSV file.
Combined with upcoming update(s) that enables locking down an entire TLD to only delegate domains with a nameserver restricted reservation type, this change will enable us to restrict domain delegation to nameservers specifically specified in the allowed nameservers list, in order to prevent malicious delegation in case the registrar for a brand TLD is compromised.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=149989330
This CL defines metrics for both premium and reserved lists, but actually uses only the reserved list metrics. The premium list metrics will be used in a future CL.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=149982639
Instead of only returning the most severe one, return all applicable ones. This is because the reserved list has grown to a list of types that are not strictly comparable but orthogonal to each other. We can no longer depend on the fact that the most severe type incorporates all properties of those beneath it. Therefore returning all of them and treat them one by one in the calling site is the correct behavior.
Due to constraint imposed in eppcom.xsd, during domain checks the response can only contain a reservation reason of fewer than 32 characters, therefore we are returning the message for the type with highest severity, in case of multiple reservation types for a label.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=149776106
This is an error condition that will soon throw an exception when
attempting to register the domain name, so it's good to let the registry
operator know of the error when it is first introduced.
Unfortunately there's still a backdoor that allows duplicate labels
that's harder to protect against (that this commit doesn't cover): the
case where reserved lists are already applied to a TLD, then one of the
reserved lists is updated to add another auth code, which then conflicts
with one on a different reserved list.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=149443007
Principally, this moves a load method into DatastoreHelper that is now
only used by tests.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=148649087
This fixes Java7 generic type inference errors in premium list code
and upgrades the Truth library to v0.32, because we're now using
assert(e).hasMessageThat() which is not in prior versions.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=148501657
It was kind of messy having all of that logic living alongside the
entities themselves.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=148498024
This also cleans up the PremiumList API so that it only has one
method for checking premium prices, which is by TLD, rather than two.
I will be refactoring a lot of the static methods currently residing in
the PremiumList class into a separate utils class, but I don't want to
include too many changes in this one CL.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=148475345
This is the first step in the migration to remove the need to load all of
the premium list entries every time the cache expires (which causes slow-
downs). Once this is deployed, we can re-save all premium lists, creating
the bloom filters, and then the next step will be to read from them to
more efficiently determine if a label might be premium.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=147525017
This allows us to use util methods from within config, which is a useful thing
to be able to do for, e.g., being able to log errors while loading configuration.
It makes sense that the util package should be at the very base of the
class inheritance hierarchy; config seems logically higher than it.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=144324273
This primarily addresses issues with TMCH testing mode and email sending utils.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=143710550
It wasn't being used by any actual code, and having helper methods handling
saving/persistence on entities like this is not a pattern we want to encourage,
since it hides Datastore transactions from further up in the call chain. The
idea is that you can always look for ofy() calls in the same layer of code to
see where persisted data is being changed.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=143036027
This defaults to null, and leaving it to null now simply disables reserved terms
exporting, rather than throwing an error every time the action runs.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=138763161
Currently EapFee is a separate class that has no inheritance from either
BaseFee and Fee. With this CL its functionality is merged into the Fee class
and the type of the fee can be identified by the FeeType enum in the Fee class.
Future custom fees can follow the same pattern.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=133627570
The backfill for these is no longer necessary as all of the Registry entities in all environments have been backfilled.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=131984138
This change replaces all Ref objects in the code with Key objects. These are
stored in datastore as the same object (raw datastore keys), so this is not
a model change.
Our best practices doc says to use Keys not Refs because:
* The .get() method obscures what's actually going on
- Much harder to visually audit the code for datastore loads
- Hard to distinguish Ref<T> get()'s from Optional get()'s and Supplier get()'s
* Implicit ofy().load() offers much less control
- Antipattern for ultimate goal of making Ofy injectable
- Can't control cache use or batch loading without making ofy() explicit anyway
* Serialization behavior is surprising and could be quite dangerous/incorrect
- Can lead to serialization errors. If it actually worked "as intended",
it would lead to a Ref<> on a serialized object being replaced upon
deserialization with a stale copy of the old value, which could potentially
break all kinds of transactional expectations
* Having both Ref<T> and Key<T> introduces extra boilerplate everywhere
- E.g. helper methods all need to have Ref and Key overloads, or you need to
call .key() to get the Key<T> for every Ref<T> you want to pass in
- Creating a Ref<T> is more cumbersome, since it doesn't have all the create()
overloads that Key<T> has, only create(Key<T>) and create(Entity) - no way to
create directly from kind+ID/name, raw Key, websafe key string, etc.
(Note that Refs are treated specially by Objectify's @Load method and Keys are not;
we don't use that feature, but it is the one advantage Refs have over Keys.)
The direct impetus for this change is that I am trying to audit our use of memcache,
and the implicit .get() calls to datastore were making that very hard.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=131965491
Also had to add an EnumParameter class to support
List<T extends Enum<T>>, as these aren't natively supported by
JCommander (although single Enum parameters are.)
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=129464699
This is one of several CLs in a sequence for allowing per-TLD DNS
implementations.
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=129445641
This is better than the previous way of using the canonical name of the class,
because the previous way did not allow for refactoring, and also required the
PremiumPricingEngine to live in the model package lest there be circular
dependencies, which does not seem ideal.
Note that, for reasons of backwards compatibility with existing persisted data,
the name of the static premium pricing engine has been set to its canonical
class name, but the class can now be refactored going forward so long as this
string remains unchanged, and any new pricing engine implementations can use
whatever string key they want (it doesn't have to be a canonical class name).
-------------
Created by MOE: https://github.com/google/moe
MOE_MIGRATED_REVID=129215185
