Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[nexus] Managing local rack -> managing local fleet #1276

Open
smklein opened this issue Jun 24, 2022 · 0 comments
Open

[nexus] Managing local rack -> managing local fleet #1276

smklein opened this issue Jun 24, 2022 · 0 comments
Labels
customer For any bug reports or feature requests tied to customer requests

Comments

@smklein
Copy link
Collaborator

smklein commented Jun 24, 2022

Some operations within Nexus are implemented as "manage the state which may exist within my local rack". This includes:

  • Awaiting handoff from RSS
  • Ensuring a rack-wide CRDB instance exists
  • Ensuring sufficient redundancy for services exists with a rack

However, longer-term, we would ideally migrate many of these operations to be "fleet-wide" instead of "rack-wide". This way, one nexus could control multiple racks simultaneously, ensure that CRDB nodes are distributed within an AZ, and ensure that service redundancy suffices for multi-rack failure scenarios.

For additional context, see: https://github.com/oxidecomputer/omicron/pull/1234/files/28d87f51ab88cce3d8ff2560a8996904e8c78f81#diff-5a93a4691987ea1b28d848375a2728abcb26cec85d477d051243cb1863198392
@smklein smklein mentioned this issue Jun 24, 2022
Merged
@smklein smklein mentioned this issue Aug 2, 2022
Merged
@smklein smklein mentioned this issue Dec 15, 2022
Merged
smklein added a commit that referenced this issue Dec 19, 2022
@smklein
[nexus] Remove project_id, rack_id from IP pools (#2056)
f47aefc 
## Before this PR

- IP Pools could exist in at most one project. IP allocation during
instance creation occurred by [either by requesting an IP pool belonging
to a project, or by "just looking for any unreserved IP
Pool"](https://github.com/oxidecomputer/omicron/blob/79765a4e3b39a29bc9940c0e4a49c4364fbcc9e3/nexus/src/db/queries/external_ip.rs#L186-L212).
As discussed in #2055 ,
our intention is for IP pools to be used across multiple projects, and
for projects to be able to use multiple IP pools.
- "Service" IP pools were indexed by rack ID, though (as documented in
#1276 ), they should
probably be accessed by AZ instead.

## This PR

- Adds a default IP pool named `default`, which is used for address
allocation unless a more specific IP pool is provided
- Removes "project ID" from IP pools (and external IP addresses)
- Removes "rack ID" from IP pool API and DB representation

## In the future

- This PR doesn't provide the many-to-many connection between projects
and IP pools that we eventually want, where projects can be configured
to use different IP pools for different purposes. However, by removing
the not-quite-accurate relationship that an IP pool must belong to a
*single* project, the API moves closer towards this direction.
- We probably should access the `service_ip_pool` API with the AZ UUID
used for the query, but since AZs don't exist in the API yet, this has
been omitted.

Part of #2055
leftwo pushed a commit that referenced this issue Apr 30, 2024
Update crucible and propolis
efc875a 
Propolis:
Update oximeter dependency to pull in automatic producer registration (#689)
Propagate ReplaceResult up; return disk status (#687)
Enable clippy warnings for lossless casts
Update rustls deps for CVE-2024-32650
migration: refrain from offering all pages when possible (#682)

Crucible:
DTrace probes for IO on/off the network (#1284)
Update oximeter dep to pull in automatic producer registration (#1279)
Remove `ReadResponse` in favor of `RawReadResponse` (#1212)
Fix typo in DTrace upstairs_info (#1276)
replace needing no work should not be an error (#1275)
Add some DTrace scripts to the package. (#1274)
More Pantry updates for Region replacement (#1269)
Send the correct task count for reconciliations (#1271)
Raw extent cleanup (#1268)
@leftwo leftwo mentioned this issue Apr 30, 2024
Merged
leftwo added a commit that referenced this issue Apr 30, 2024
@leftwo
Update crucible and propolis (#5673)
fcf7980 
Propolis:
Update oximeter dependency to pull in automatic producer registration (#689)
Propagate ReplaceResult up; return disk status (#687)
Enable clippy warnings for lossless casts
Update rustls deps for CVE-2024-32650
migration: refrain from offering all pages when possible (#682)

Crucible:
DTrace probes for IO on/off the network (#1284)
Update oximeter dep to pull in automatic producer registration (#1279)
Remove `ReadResponse` in favor of `RawReadResponse` (#1212)
Fix typo in DTrace upstairs_info (#1276)
replace needing no work should not be an error (#1275)
Add some DTrace scripts to the package. (#1274)
More Pantry updates for Region replacement (#1269)
Send the correct task count for reconciliations (#1271)
Raw extent cleanup (#1268)

---------

Co-authored-by: Alan Hanson <alan@oxide.computer>
@twinfees twinfees added the customer For any bug reports or feature requests tied to customer requests label Nov 1, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
customer For any bug reports or feature requests tied to customer requests
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@smklein @twinfees