[SERVER-6108] Error message 13257 could use clarification Created: 15/Jun/12  Updated: 07/Apr/13  Resolved: 07/Apr/13

Status: Closed
Project: Core Server
Component/s: Replication
Affects Version/s: 2.0.5
Fix Version/s: None

Type: Bug Priority: Trivial - P5
Reporter: Jed Smith Assignee: Scott Hernandez (Inactive)
Resolution: Incomplete Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Operating System: ALL
Participants:

 Description   

Stupid simple request. I hesitate to call it a bug, but I also hesitate to call it an improvement.

I made a mistake deploying a Mongo node yesterday and it ran out of space while allocating its journal and local databases. Before I had realized it, I tried adding it to a replica set and got this errmsg in the set status:

syncThread: 13257 cmdline oplogsize (5210) different than existing (2046) see: http://dochub.mongodb.org/core/increase-oplog"

That confused the hell out of us. It makes sense once you understand, but it launched us down a red herring hole. It's just this replica set error message that was a little strange at first glance, because we thought we had misconfigured the oplog somehow. Again, I should have realized sooner that I had ran the box out of space, and I'm sure there's something in the log about it, but seeing that in replica set status was just confusing.

Does Mongo consider the local database stable even if it runs out of space while making it? Could it instead say, here, "I couldn't fully populate local, wtf?" or something? That's an old URI, too, which redirects elsewhere ... and the instructions aren't terribly appropriate for the situation at hand.

Like I said, stupidly trivial.



 Comments   
Comment by Scott Hernandez (Inactive) [ 07/Apr/13 ]

Yep, logs would have been useful and without them it is hard to guess.

If you started with files, instead of with an empty dbpath, then the oplog could have been from somewhere else. But guessing is not always so helpful

Comment by Jed Smith [ 16/Jun/12 ]

I went to look, and we failed to capture it. We've been having issues with our logging coming up correctly on new nodes and we only became aware recently, so that was one of the affected, unfortunately. Disservice to you, I know.

This was definitely the sequence of events:

  • Create instance, add Mongo, start Mongo.
  • Mongo began preallocating its journal (we patch to force preallocation) and creating local (maybe?).
  • Later, I added it to the replica set in question without noticing the disk space issue.
  • I got this message in replica set status on the primary.

(I say maybe because I don't know if local creation happens before or after replica set configuration when it's started with --replSet.)

The way our config works, the oplog size is specified on the command line and it is hardcoded into the Supervisor config. We had changed it from 10 GB to 5 GB, but that was days before we deployed this instance. There's no way Mongo ever got a value other than 5210. If the message should be unrelated, that's really odd and I wish I had logs.

Comment by Scott Hernandez (Inactive) [ 16/Jun/12 ]

This message just means that you have an oplog size in your config/cmdLine which is different the currently allocated one – probably indicating you changed the value and want to resize the oplog.

If you had problems running out of space during the initial oplog/local-db creation that would be unrelated to this message. Can you post the logs for that out of disk space event?

Generated at Thu Feb 08 03:10:46 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.