[SERVER-20601] System unusuable during mongodump from WiredTiger Created: 24/Sep/15 Updated: 06/Apr/23 Resolved: 23/Oct/15 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | Performance, Tools, WiredTiger |
| Affects Version/s: | 3.0.5 |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Minor - P4 |
| Reporter: | Scott Lowe | Assignee: | David Hows |
| Resolution: | Incomplete | Votes: | 0 |
| Labels: | needs-repo | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Operating System: | ALL |
| Steps To Reproduce: |
|
| Participants: |
| Description |
|
Background: We just moved our production system from MMAP to WiredTiger, and are generally seeing extremely poor performance.
I thought that maybe this was something specific to our workload that might be causing problems. I was going to change some of the read-intensive collections from snappy compression, to no compression, to see if that improved things at all, however today I came across the mongodump issue, which is a show stopper for us (as we use mongodump for live backups). -------------------------------------- In a nutshell: When mongodump is running, all application queries to the db seem to be blocked. (It's also worth noting that mongodump takes a lot longer to run on WT, which though not a major problem, does compound the issue). |
| Comments |
| Comment by Ramon Fernandez Marina [ 23/Oct/15 ] | ||||
|
sosh, we haven't heard back from you for some time so we're going to close this ticket. If this is still an issue for you please provide the information requested above by David so we can investigate further. Thanks, | ||||
| Comment by Alexander Gorrod [ 12/Oct/15 ] | ||||
|
Hi bgsosh Have you had a chance to gather the information? Without further information there is very little else we can do to resolve your issue. | ||||
| Comment by Scott Lowe [ 02/Oct/15 ] | ||||
|
Hi, Sorry for the delay, I must not have noticed the email alert about your comment. Unfortunately we've had to revert back so MMAP (it was a production system, so we had very limited time in which to either fix or revert). However, other than the storage engine change, everything else has remained the same, so perhaps the output form the diagnostic script will still be useful - I'll try and run that today. Thanks | ||||
| Comment by David Hows [ 28/Sep/15 ] | ||||
|
Hi bg, To begin understanding this issue we will need a lot more data to see what is going on within your system. Would you be able to run the following commands on your system while it runs normally, then start the mongodump task. This will give some insight into whats going on at the time.
Additionally, can you please run the MongoDB general system diagnostic script mdiag to gather some basic information about your system, much of what is included here will be for context on the outputs of the commands above.
When you're finished please upload the three files (mdiag output, iostat output and serverstatus output) to the ticket. Thanks, |