[SERVER-38681] Make it possible to have a TaskExecutor and ConnectionPool that don't auth Created: 17/Dec/18 Updated: 06/Dec/22 Resolved: 12/Nov/21 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Networking |
| Affects Version/s: | None |
| Fix Version/s: | features we're not sure of |
| Type: | Improvement | Priority: | Major - P3 |
| Reporter: | Mathias Stearn | Assignee: | Backlog - Service Architecture |
| Resolution: | Won't Do | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||
| Assigned Teams: |
Service Arch
|
||||||||
| Participants: | |||||||||
| Description |
|
The ReplicaSetMonitor should be running its isMaster calls on un-authed connections, but this isn't currently possible with a TaskExecutor. |
| Comments |
| Comment by Lauren Lewis (Inactive) [ 12/Nov/21 ] |
|
We haven’t heard back from you for some time, so I’m going to close this ticket. If this is still an issue for you, please provide additional information and we will reopen the ticket. |
| Comment by Andy Schwerin [ 26/Dec/18 ] |
|
I think putting the justification into the description would suffice. I'm interested further in the future in reducing the number of round trips to the first fully authenticated RPC in the common case, both to handle connection scaling and because I still believe in the dream of |
| Comment by Mira Carey [ 20/Dec/18 ] |
|
Because it has the option to (because the drivers do so), and because waiting for auth'd connections adds overhead in targeting that doesn't deliver any extra user value. The pathological case is a replset with a newly elected primary on the opposite side of the earth (from our mongos). We're looking at 2x the round trips, so 2x the wait before we start establishing connections and servicing requests. Round trips:
I'd like to make it possible for the rsm to start fielding requests after the initial isMaster. I can see a weakening of the ticket description to "should be running its isMaster calls as soon as possible, including on new un-authd connections if required". would that work for you? (probably by extending the current isMaster hook to update the rsm) |
| Comment by Andy Schwerin [ 18/Dec/18 ] |
|
I'd like some clarity on why we want this. I've had it explained to me before, but I'd like to see it written down. Specifically, why should the replica set monitor run its ismaster calls on un-authed connections? |