[SERVER-29867] moveChunk command ignores "_waitForDelete:" option Created: 26/Jun/17 Updated: 30/Oct/23 Resolved: 29/Aug/17 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Sharding |
| Affects Version/s: | 3.5.9 |
| Fix Version/s: | 3.4.7 |
| Type: | Bug | Priority: | Major - P3 |
| Reporter: | Nathan Myers | Assignee: | Esha Maharishi (Inactive) |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||
| Backwards Compatibility: | Fully Compatible | ||||||||
| Operating System: | ALL | ||||||||
| Backport Requested: |
v3.4
|
||||||||
| Sprint: | Sharding 2017-07-31 | ||||||||
| Participants: | |||||||||
| Linked BF Score: | 0 | ||||||||
| Description |
|
The moveChunk command offers a "waitForDelete" option that does the obvious. It ignores "_waitForDelete" spelled without the leading underscore. It should treat both spellings identically. The documentation should be changed to suggest the non-underscore version. (We will support the underscore spelling indefinitely, for backward compatibility.) Drivers should also accept the non-underscore option, and corresponding IDE annotations should suggest it. |
| Comments |
| Comment by Esha Maharishi (Inactive) [ 21/Jul/17 ] |
|
Author: {u'username': u'EshaMaharishi', u'name': u'Esha Maharishi', u'email': u'esha.maharishi@mongodb.com'}Message: |
| Comment by Nathan Myers [ 21/Jul/17 ] |
|
Probably. But _waitForDelete makes less difference in |
| Comment by Kaloian Manassiev [ 21/Jul/17 ] |
|
Is this a problem in 3.4 as well? |
| Comment by Esha Maharishi (Inactive) [ 20/Jul/17 ] |
|
|
| Comment by Esha Maharishi (Inactive) [ 10/Jul/17 ] |
|
Also noting that mongos only passes '_waitForDelete' (with underscore) to the config server: https://github.com/mongodb/mongo/blob/r3.5.9/src/mongo/s/commands/cluster_move_chunk_cmd.cpp#L199 |
| Comment by Esha Maharishi (Inactive) [ 07/Jul/17 ] |
|
nathan.myers, I think this might be backwards - shards respect "waitForDelete" but not "_waitForDelete"? |
| Comment by Nathan Myers [ 26/Jun/17 ] |
|
The moveChunk command will now interpret a "waitForDelete:" option identically as "_waitForDelete", instead of ignoring it. Note that, starting in 3.6, if the waitForDelete option is not chosen, the command returns immediately after the chunk is copied, but the server waits before beginning to delete the orphan chunk. The waiting period is determined by the orphanCleanupDelaySecs server parameter, which defaults to 15 minutes. (This delay is intended to allow queries still running on secondary servers, which may depend on documents in the chunk, to complete before deletion begins.) |