[SERVER-21] Feature-Req: Flush this record(s) to disk Created: 17/Apr/09 Updated: 25/Aug/09 Resolved: 25/Aug/09 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Stability |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | New Feature | Priority: | Minor - P4 |
| Reporter: | Jim Jones | Assignee: | Eliot Horowitz (Inactive) |
| Resolution: | Duplicate | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Participants: |
| Description |
|
It would be interesting if a client could issue requests for selective msync()s on the server. |
| Comments |
| Comment by Eliot Horowitz (Inactive) [ 25/Aug/09 ] |
|
see |
| Comment by Jim Jones [ 01/Jun/09 ] |
|
Well, mongodump is still a lot more expensive than a constant-time snapshot on our SAN. With mongodump we have to:
Both can take a while for multi-gigabyte databases. With a snapshot we'd only have to:
I realize we could just stop mongod altogether but losing all the cached data is undesirable, we'd like to take snapshots hourly. |
| Comment by Eliot Horowitz (Inactive) [ 31/May/09 ] |
|
unless you need an exaact moment in time snapshot - using mongodump will be almost as fast - and won't halt the db |
| Comment by Jim Jones [ 31/May/09 ] |
|
A related use-case: For backup purposes it would be good if we could, on request, sync the entire datastore to disk and halt mongod temporarily (for taking a filesystem snapshot). This could either be implemented through client commands, or under unix with the traditional SIGSTOP/SIGCONT signals. Ideally mongod would also report back after entering the frozen state, e.g. by touching a predefined file (configured by command line switch). A simple backup-script could then look like this: kill -SIGSTOP `pidof mongod`
|