[SERVER-15852] replication oplog stream goes back in time under load Created: 28/Oct/14  Updated: 10/Dec/14  Resolved: 29/Oct/14

Status: Closed
Project: Core Server
Component/s: Replication
Affects Version/s: 2.7.8
Fix Version/s: None

Type: Bug Priority: Major - P3
Reporter: Davide Italiano Assignee: Scott Hernandez (Inactive)
Resolution: Done Votes: 0
Labels: 28qa
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Related
Operating System: ALL
Participants:

 Description   

Doesn't happen in 2.6.

Upstream version:

commit 0e9cb3b20498b9f167afaff7a5c4a4d1da7e06a2
Author: Dan Pasette <dan@10gen.com>
Date:   Wed Oct 22 14:10:37 2014 -0400
 
    BUMP 2.7.8

Relevant bits in the log:

$ tail -n 20 repl2.txt.2014-10-28T23-17-32 
2014-10-28T23:17:31.731+0000 I STORAGE  [FileAllocator] done allocating datafile /home/ec2-user//mongo/repl2/thisdb10.0, size: 64MB,  took 0.002 secs
2014-10-28T23:17:31.733+0000 I INDEXING [rsSync] build index on: thisdb10.mycoll10 properties: { v: 1, key: { _id: 1 }, name: "_id_", ns: "thisdb10.mycoll10" }
2014-10-28T23:17:31.734+0000 I INDEXING [rsSync]         building index using bulk method
2014-10-28T23:17:31.734+0000 I INDEXING [rsSync] build index done.  scanned 11 total records. 0 secs
2014-10-28T23:17:31.734+0000 I REPLSETS [rsSync] initial sync cloning db: thisdb15
2014-10-28T23:17:31.735+0000 I INDEXING [rsSync] allocating new ns file /home/ec2-user//mongo/repl2/thisdb15.ns, filling with zeroes...
2014-10-28T23:17:31.901+0000 I STORAGE  [FileAllocator] allocating new datafile /home/ec2-user//mongo/repl2/thisdb15.0, filling with zeroes...
2014-10-28T23:17:31.906+0000 I STORAGE  [FileAllocator] done allocating datafile /home/ec2-user//mongo/repl2/thisdb15.0, size: 64MB,  took 0.005 secs
2014-10-28T23:17:31.910+0000 I INDEXING [rsSync] build index on: thisdb15.mycoll15 properties: { v: 1, key: { _id: 1 }, name: "_id_", ns: "thisdb15.mycoll15" }
2014-10-28T23:17:31.910+0000 I INDEXING [rsSync]         building index using bulk method
2014-10-28T23:17:31.910+0000 I INDEXING [rsSync] build index done.  scanned 79 total records. 0 secs
2014-10-28T23:17:31.910+0000 I REPLSETS [rsSync] initial sync data copy, starting syncup
2014-10-28T23:17:31.910+0000 I REPLSETS [rsSync] oplog sync 1 of 3
2014-10-28T23:17:32.373+0000 F REPLSETS [rsSync] replication oplog stream went back in time. previous timestamp: 54502408:4765 newest timestamp: 545023e6:10eb
2014-10-28T23:17:32.373+0000 I -        [rsSync] Fatal Assertion 18905
2014-10-28T23:17:32.373+0000 I -        [rsSync] 
 
***aborting after fassert() failure



 Comments   
Comment by Davide Italiano [ 29/Oct/14 ]

I think this is the same root cause – can't repro in master.

Comment by Spencer Brody (Inactive) [ 29/Oct/14 ]

I wonder if this is the same root cause as SERVER-15770. I could repro SERVER-15770 a week or two ago, but no longer can. davide.italiano, can you repro this on master?

Generated at Thu Feb 08 03:39:11 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.