[SERVER-4663] mongodump performance Created: 11/Jan/12 Updated: 15/Aug/12 Resolved: 11/Apr/12 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Replication |
| Affects Version/s: | 2.0.1 |
| Fix Version/s: | None |
| Type: | Question | Priority: | Major - P3 |
| Reporter: | Theo Carpenter | Assignee: | Scott Hernandez (Inactive) |
| Resolution: | Incomplete | Votes: | 0 |
| Labels: | backup, replication | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Environment: |
EC2 2.4 server replication member on amazon Linux AMI. |
||
| Participants: |
| Description |
|
replication member is using mongodump command to backup database collections one at time and then transport them to another site for DR. The mongodump process takes so long that it places the replication server hours behind in replication. Replication doesn't catch up before the next night's backup kicks off, which creates a slowdown during the backup until disk activity maxes out and backups and replication grind to a halt. What can we do to improve dump performance? |
| Comments |
| Comment by Scott Hernandez (Inactive) [ 12/Jan/12 ] |
|
Can you try using the --forceTableScan flag? http://www.mongodb.org/display/DOCS/Import+Export+Tools#ImportExportTools-PerformanceTips What does mongostat/iostat -xm 2 look like during the backup process? If you have less then 1TB of data, and replication can keep up, then you can use a single EBS on a special backup replica which is hidden to allow you to use EBS snapshots without having to synchronize the process across multiple EBS volumes. What is your disk configuration now? |