[SERVER-1714] multiplex mongos -> mongod connecions after successful read op Created: 01/Sep/10 Updated: 12/Jul/16 Resolved: 08/Jan/14 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Sharding |
| Affects Version/s: | None |
| Fix Version/s: | 2.5.5 |
| Type: | Improvement | Priority: | Major - P3 |
| Reporter: | Eliot Horowitz (Inactive) | Assignee: | Greg Studer |
| Resolution: | Done | Votes: | 10 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||||||||||||||||||||||||||||||||||||||
| Participants: | |||||||||||||||||||||||||||||||||||||||||||||
| Description |
|
Can't do this until state moves from per connection to per reqeust. |
| Comments |
| Comment by Grégoire Seux [ 28/Feb/13 ] |
|
We have exactly the same issue: |
| Comment by Kevin J. Rice [ 27/Feb/13 ] |
|
We're seeing (with 48 shards in our cluster, 24 primaries on each of two boxes, primary/replica/arbiter setup w/ 3 configs), upwards of 600+ connections per primary mongod process. This is huge. It prevents us from going to rather large numbers of shards, say 500 on one box. Further, IIRC, if I have 10 processes connect to one mongos and that mongos connects to 4 shards, I've really instantiated 40 connections. This is huge. I am going to have 3000 servers, each with a pymongo client, connecting and sending data. This means 3000 * 4 = 12,000 connections. I may be able to handle this with virtual IP addresses for each of the mongod's. But, it's an ugly workaround. |
| Comment by Eliot Horowitz (Inactive) [ 27/Jul/11 ] |
|
with stack size changes, this is less important |
| Comment by auto [ 19/May/11 ] |
|
Author: {u'login': u'erh', u'name': u'Eliot Horowitz', u'email': u'eliot@10gen.com'}Message: don't unset sharding state when a shard connection is done |
| Comment by Eliot Horowitz (Inactive) [ 18/May/11 ] |
|
Left:
|
| Comment by Eliot Horowitz (Inactive) [ 11/Nov/10 ] |
|
this is going to be too expensive till we can re-work network so version is per request |