This change adds a new column to the user table, which will hold a
random code for account activation. And another to hold the signup
request ID as another layer of verification.
Issue: #6428
Change-Id: Icd46cb5d8fc76102264d599aca27686cd8b2e84e
Currently we have large set of nodes that are already disqualified and
we are not sending bloom filters to them. The issue is that we are still
generating filters for them while garbage collection process. Even if
we have only segment with one piece which was stored on this node. This
consumes additional memory and processing powers.
This change is changing logic behind `AllPieceCounts` (renamed to
ActiveNodesPieceCounts) to return piece count for all nodes except disqualified one (even with piece count = 0). With this change we can
modify GC observer to skip nodes that where not returned by
ActiveNodesPieceCounts.
Change-Id: Ic75159135abe535084d8aeee560bb801a4a03e17
This change gives the proper type to TS class fields generated from
nillable Go struct fields. Previously, Go struct fields having a nil
representation ([]Type, *Type, etc.) were translated into TypeScript as
"Type | undefined". This isn't correct because these fields, when nil,
are given the value "null" when marshalled into JSON. This change fixes
this issue by giving these fields the type "Type | null".
Change-Id: I5a1a83eb3810a3cba10895bb2f0f75ca5fd7d1b5
the new bus_factor calculation doesn't make sense with different classes, as we have overlaps.
For example: it can detect a risk if we loose one country and one different subnet (with possible overlap).
It's better to calculate the stat and bus_factor per class (net, country, ...).
It also makes it easier to measure execution time per class.
Change-Id: I7d4d5f7cb811cd50c5831077b43e001908aab96b
Run Versions test is intermittent and slow.
We need to rewrite it with storj-up, preferable with a container where
old versions are cached, instead of rebuilding them everytime
(which is -- IMHO -- the root cause of the intermittency).
As we discussed on the chat, we can separate the runs, until we prioritize this work.
Change-Id: I7fdf6fdb20625fd76d6334be0d0afe72af1b734a
For S3 compatibility we should not allow returning pure delete
marker object. If metabase returns delete marker metainfo will
return MethodNotAllowed rpc status.
https://github.com/storj/storj/issues/6522
Change-Id: I89804b2bd22da0e5beec8f106e74b74733e19a52
This change fixes an issue that caused the Papa Parse worker plugin to
break during Vite's watch mode.
Change-Id: I84d03d38e2d0d6fe4b7536706638640d3ed259a6
Show buckets limit card on project dashboard instead of coupon card if user is in paid tier or is not the owner of currently selected project.
Issue:
https://github.com/storj/storj/issues/6492
Change-Id: I033bbcced1b0fb77436607847cd9f10830330009
This reverts commit 2b4f347c33.
Reason for revert: not a zero-downtime migration. We need to deploy the dbx changes before completely dropping the column.
Change-Id: Ic9215650242b7848d54f8a5f863a13bc18b60149
This change adds two new admin endpoints to freeze users for legal
review and to remove them from that state
Issue: storj/storj-private#492
Change-Id: I6c8e3ffcb80375e81e78bc6ecc785c1047328cf7
This change makes the TypeScript API code generator properly handle
struct fields with the "omitempty" option in the JSON struct tag.
Change-Id: I9b22ce33a8b8c39c115ec827a8e5b7e85d856f83
Not partnered users were redirected to onboarding pricing plan step and back to all projects dashboard in case when they have not finished onboarding. That caused tons of redundant API requests to be done.
Change-Id: I454a8cc29730943b84a351416b78c4281c47f6ca
This change adds an endpoint to the back office API that returns
placement IDs and their locations.
References #6503
Change-Id: I20ee1c82dcb647d6d264317beceeb5e70f7a8e87
I made several commits to add support to the API generator for
anonymous types because the example was using them.
Later, I was told that we don't need to support them, so the example
should have never used anonymous types.
Supporting anonymous types has added complexity to the API generator and
we are finding bugs in corner cases whose fixes lead to adding more
complexity and some fixes doesn't seem to have a neat solution without
doing a big refactoring.
We decided to reduce the side effects that supporting anonymous type has
brought just removing it.
This commit remove the support for anonymous types and reflect that in
the example.
Change-Id: I2f8c87a0db0e229971ab1bef46cca16fee924191
Implement the authorization that will hook into each endpoint handler
through a wrapping handler for defining the permissions that each
endpoint requires.
Change-Id: I9c8f12b58f48e849e7ea35f372dddce5c9cfc5b5
For convenience of not having to modify the API generator to contemplate
the path prefix that we are adding to the back office server, we define
the path prefix in a constant than the admin server and the definition
of the API uses to adapt the router and the generated code.
Change-Id: Ic557b0e6e88e930e03647835759bb34e06e8bb48
This change fixes an incorrect invocation of the URL object constructor
in generated TypeScript HTTP clients.
Change-Id: I9011bc535f2096374d20b74b401d4cc38a0451fb
We decided that we won't use seprate table for handling pending
objects. We need to remove related code.
https://github.com/storj/storj/issues/6421
Change-Id: I442b0f58da75409f725e08e2cd83d29ed4f91ec6
Removed MetadataSize, RepairEgress, AuditEgress fields from usage report because they are not relevant.
Issue:
https://github.com/storj/storj/issues/6498
Change-Id: I8ef7d56bd1b01cdb53cec8c67dd47a6a0e7fa184
This change prevents Go code produced by the API generator from
importing its own package. Previously, we tried to prevent self imports
by skipping import paths whose last segment matched the generated Go
code's package name. However, aliased imports circumvented this.
We now require API definitions to define the Go package path so that we
can compare this with the import path directly.
Change-Id: I7ae7ec5e1a342d2f76cd28ff72d4cd7285c2820a
This change wraps account freeze code in DB transactions to prevent
freeze inconsistencies resulting from errors that happen in the process
of freezing accounts.
Change-Id: Ib67fb30dc33248413d3057ceeac5c2f410f551d5
This change makes the example API definition use a hardcoded timestamp
rather than the current time so that files aren't unnecessarily
changed each time the API code is generated.
Change-Id: I72e4ec98d29345b9f16ca2ca38a91e593f83ea87
To be compatible with S3 we need to return 'Method Not Allowed' when
delete marker is requested and to do this libuplnk needs to know about
delete marker. It will be returned only if object version will be
specified.
Change-Id: I288da5566c74e1b4951f7cd249dbf34622b92e91
We decided that we won't use seprate table for handling pending
objects. We need to remove related code.
After this change we still needs to remove related code from metabase.
https://github.com/storj/storj/issues/6421
Change-Id: Idaaa8ae08b80bab7114e316849994be171daf948
Papa Parse, the library we use to parse CSV files in the satellite UI,
uses a blob URL for its worker. This isn't allowed by our content
security policy, so this change implements a Vite plugin that writes
the worker code to a file that is statically served.
Change-Id: I0ce58c37b86953a71b7433b789b72fbd8ede313d
PathPrefix is the full path of the subrouter passed to the back office
server when it hooks into another server, in this case, the satellite
admin server.
PathPrefix allows to serve the static assets from the root of the
sub-router when the prefix is stripped before accessing them.
There was a bug where the PathPrefix weren't used and a hard-coded path
was used.
Test passed because the back-office server is hooked into the satellite
admin server with a subrouter with path `/back-office/` which matched
the hard-code value, however, it wouldn't work if that path changed or
it is hooked into another server with a different subrouter path,
despite it was set to PathPrefix.
This commit fixes that bug.
Change-Id: Id4a0d86329eb563b008b3fc6f8eb7b51cbfd2e6f
