MongoDB: mongoimport loses connection when importing big files - mongodb

I have some trouble importing a JSON file to a local MongoDB instance. The JSON was generated using mongoexport and looks like this. No arrays, no hardcore nesting:
{"_created":{"$date":"2015-10-20T12:46:25.000Z"},"_etag":"7fab35685eea8d8097656092961d3a9cfe46ffbc","_id":{"$oid":"562637a14e0c9836e0821a5e"},"_updated":{"$date":"2015-10-20T12:46:25.000Z"},"body":"base64 encoded string","sender":"mail#mail.com","type":"answer"}
{"_created":{"$date":"2015-10-20T12:46:25.000Z"},"_etag":"7fab35685eea8d8097656092961d3a9cfe46ffbc","_id":{"$oid":"562637a14e0c9836e0821a5e"},"_updated":{"$date":"2015-10-20T12:46:25.000Z"},"body":"base64 encoded string","sender":"mail#mail.com","type":"answer"}
If I import a 9MB file with ~300 rows, there is no problem:
[stekhn latest]$ mongoimport -d mietscraping -c mails mails-small.json
2015-11-02T10:03:11.353+0100 connected to: localhost
2015-11-02T10:03:11.372+0100 imported 240 documents
But if try to import a 32MB file with ~1300 rows, the import fails:
[stekhn latest]$ mongoimport -d mietscraping -c mails mails.json
2015-11-02T10:05:25.228+0100 connected to: localhost
2015-11-02T10:05:25.735+0100 error inserting documents: lost connection to server
2015-11-02T10:05:25.735+0100 Failed: lost connection to server
2015-11-02T10:05:25.735+0100 imported 0 documents
Here is the log:
2015-11-02T11:53:04.146+0100 I NETWORK [initandlisten] connection accepted from 127.0.0.1:45237 #21 (6 connections now open)
2015-11-02T11:53:04.532+0100 I - [conn21] Assertion: 10334:BSONObj size: 23592351 (0x167FD9F) is invalid. Size must be between 0 and 16793600(16MB) First element: insert: "mails"
2015-11-02T11:53:04.536+0100 I NETWORK [conn21] AssertionException handling request, closing client connection: 10334 BSONObj size: 23592351 (0x167FD9F) is invalid. Size must be between 0 and 16793600(16MB) First element: insert: "mails"
I've heard about the 16MB limit for BSON documents before, but since no row in my JSON file is bigger than 16MB, this shouldn't be a problem, right? When I do the exact same (32MB) import one my local computer, everything works fine.
Any ideas what could cause this weird behaviour?

I guess the problem is about performance, any way you can solved used:
you can use mongoimport option -j. Try increment if not work with 4. i.e, 4,8,16, depend of the number of core you have in your cpu.
mongoimport --help
-j, --numInsertionWorkers= number of insert operations to run
concurrently (defaults to 1)
mongoimport -d mietscraping -c mails -j 4 < mails.json
or you can split the file and import all files.
I hope this help you.
looking a little more, is a bug in some version
https://jira.mongodb.org/browse/TOOLS-939
here another solution you can change the batchSize, for default is 10000, reduce the value and test:
mongoimport -d mietscraping -c mails < mails.json --batchSize 1

Quite old, but I struggled on same issue.
If you want to import big files, especially remote with Compass or by Program just add
&wtimeoutMS=0
to your Connection-String. This removes Timeout on Write-Operations.

Related

ERROR: child process failed, exited with error number 51 MongoDB

Getting this error while restarting MongoDB , I am using Mongo 3.2.4 and doing this set up on a new machine
Starting mongod... about to fork child process, waiting until server is ready for connections.
forked process: 19438
ERROR: child process failed, exited with error number 51
mongod(_ZN5mongo19MmapV1ExtentManager4initEPNS_16OperationContextE+0x4A8) [0x1040278]
mongod(_ZN5mongo26MMAPV1DatabaseCatalogEntryC1EPNS_16OperationContextENS_10StringDataES3_bb+0x187) [0x1036dc7]
mongod(_ZN5mongo12MMAPV1Engine23getDatabaseCatalogEntryEPNS_16OperationContextENS_10StringDataE+0x14E) [0x103a1de]
mongod(_ZN5mongo14DatabaseHolder6openDbEPNS_16OperationContextENS_10StringDataEPb+0x133) [0xac92a3]
----- END BACKTRACE -----
For me this error occured due to incorrect ownership of some files in my data directory. I fixed this using the following command:
sudo chown -R mongodb: /path/to/db/directory
Where mongodb was the database user in my case.
This is resolved by inserting the following lines into /etc/security/limits.conf:
mongodb soft nofile 65535
mongodb hard nofile 90000
mongodb soft nproc 65535
mongodb hard nproc 90000
We need to add the user account used to run the Mongo service. Generally, it is the mongodb user.
In my case the problem was that I haven't created the appropriate folders specified in the config files.

Connection error while mongorestore

I was trying to import some data in mongodb using :
mongoimport -d blog -c posts < posts.json
But start to get this error :
2015-11-06T11:37:52.566+0530 connected to: localhost
2015-11-06T11:37:53.159+0530 error inserting documents: lost connection to server
2015-11-06T11:37:53.159+0530 Failed: lost connection to server
2015-11-06T11:37:53.159+0530 imported 0 documents
Any help would be appreciated .
try adding a -j parameter
mongoimport -d blog -c posts -j 4 < posts.json
If this does not work, try changing the j value to 8.
This post explains it:
MongoDB: mongoimport loses connection when importing big files
I also faced the same issue. I just added --batchSize 1 option. It worked like a charm.
mongoimport -d blog -c posts -j 4 < posts.json --batchSize 1

mongorestore failing because of DocTooLargeForCapped error

I'm trying to restore a collection like so:
$ mongorestore --verbose --db MY_DB --collection MY_COLLECTION /path/to/MY_COLLECTION.bson --port 1234 --noOptionsRestore
Here's the error output (timestamps removed):
using write concern: w='majority', j=false, fsync=false, wtimeout=0
checking for collection data in /path/to/MY_COLLECTION.bson
found metadata for collection at /path/to/MY_COLLECTION.metadata.json
reading metadata file from /path/to/MY_COLLECTION.metadata.json
skipping options restoration
restoring MY_DB.MY_COLLECTION from file /path/to/MY_COLLECTION.bson
file /path/to/MY_COLLECTION.bson is 241330 bytes
error: write to oplog failed: DocTooLargeForCapped document doesn't fit in capped collection. size: 116 storageSize:1206976512 # 28575
error: write to oplog failed: DocTooLargeForCapped document doesn't fit in capped collection. size: 116 storageSize:1206976512 # 28575
restoring indexes for collection MY_DB.MY_COLLECTION from metadata
Failed: restore error: MY_DB.MY_COLLECTION: error creating indexes for MY_DB.MY_COLLECTION: createIndex error: exception: write to oplog failed: DocTooLargeForCapped document doesn't fit in capped collection. size: 116 storageSize:1206976512 # 28575
The result of the restore is a database and collection with correct names but no documents.
OS: Ubuntu 14.04 running on Azure VM.
I just solved my own problem. See answer below.
The problem seemed to be that I was using mongod on the replica set PRIMARY member.
Once I commented out the following line in /etc/mongod.conf, it worked without problems:
replSet=REPL_SET_NAME --> #replSet=REPL_SET_NAME
I assume passing the correct replica set name to the mongorestore command (like in this question) could also work, but haven't tried that yet.

MongoDB still shows empty collections after restoring from dump

After mongodump, I did mongorestore which seemed to work fine
heathers-air:db heathercohen$ mongorestore -v -host localhost:27017
2015-02-06T11:22:40.027-0800 creating new connection to:localhost:27017
2015-02-06T11:22:40.028-0800 [ConnectBG] BackgroundJob starting: ConnectBG
2015-02-06T11:22:40.028-0800 connected to server localhost:27017 (127.0.0.1)
2015-02-06T11:22:40.028-0800 connected connection!
connected to: localhost:27017
2015-02-06T11:22:40.030-0800 dump/langs.bson
2015-02-06T11:22:40.030-0800 going into namespace [dump.langs]
Restoring to dump.langs without dropping. Restored data will be inserted without raising errors; check your server log
file dump/langs.bson empty, skipping
2015-02-06T11:22:40.030-0800 Creating index: { key: { _id: 1 }, name: "_id_", ns: "dump.langs" }
2015-02-06T11:22:40.031-0800 dump/tweets.bson
2015-02-06T11:22:40.031-0800 going into namespace [dump.tweets]
Restoring to dump.tweets without dropping. Restored data will be inserted without raising errors; check your server log
file size: 4877899
30597 objects found
2015-02-06T11:22:41.883-0800 Creating index: { key: { _id: 1 }, name: "_id_", ns: "dump.tweets" }
When I try to access the data though, it's still empty and the way it looked before restore:
> show dbs
admin (empty)
dump 0.078GB
local 0.078GB
tweets (empty)
twitter (empty)
It says it found 30597 objects, where did they go?
They went into the dump database, and then into the collections dump.tweets and dump.langs. The fact that the files are contained in the folder dump means that mongorestore thinks they should be restored to the database dump (it is inferred from the path). The verbose output even explicitly states that the data is being placed into dump.langs and dump.tweets specifically.
If you specify the database you wish to restore to (with -d) and restore the specific files you will be able to restore the documents to the database you desire. Or, you can simply have a look in the dump database by running:
use dump;
db.tweets.find();
db.langs.find();

Mongodump and mongorestore; field not found

I'm trying to dump a database from another server (this works fine), then restore it on a new server (this does not work fine).
I first run:
mongodump --host -d
This creates a folder dump/db which contains all of the bson documents.
Then in the dump folder, I'm running:
mongorestore -d dbname db
This works and iterates through the files, but I get this error on dbname.system.users
Wed May 23 02:08:05 { key: { _id: 1 }, ns: "dbname.system.users", name: "_id_" }
Error creating index dbname.system.usersassertion: 13111 field not found, expected type 16
Any ideas how to resolve this?
If it realy different versions, use --noIndexRestore option. And create all index after that.
Any chance the source and destination are different versions?
In any case, to get around this, restore the collections individually using the -c flag to the target DB and then build the indexes afterward. The system collection is the one used for indexes, so it is fairly easy to recreate - try it last once everything else has been restore, and if it still fails you can always just recreate the relevant indexes.
The issue could also caused by this bug in older versions of Mongo (In my case it was 2.0.8):
https://jira.mongodb.org/browse/SERVER-7181
Basically, you get 13111 field not found, expected type 16 error when it should actually be prompting you to enter your authentication details.
And example of how I fixed it:
root#precise64:/# mongorestore /backups/demand/ondemand.05-24-2013T114223/
connected to: 127.0.0.1
[REDACTED]
Fri May 24 11:48:15 going into namespace [test.system.indexes]
Fri May 24 11:48:15 { key: { _id: 1 }, ns: "test.system.users", name: "_id_" }
Error creating index test.system.usersassertion: 13111 field not found, expected type 16
# Error when not giving username and password
root#precise64:/# mongorestore -u fakeuser -p fakepassword /backups/demand/ondemand.05-24-2013T114223/
connected to: 127.0.0.1
[REDACTED]
Fri May 24 11:57:11 /backups/demand/ondemand.05-24-2013T114223/test/system.users.bson
Fri May 24 11:57:11 going into namespace [test.system.users]
1 objects found
# Works fine when giving username and password! :)
Hope that helps anyone who's issue doesn't get fixed by the previous 2 replies!
This can also happen if you are trying to mongorestore into MongoDB 2.6+ and the dump you are trying to restore contains a system.users table in any database other than admin. In MongoDB 2.2 and 2.4 the system.userscollections could occur in any database. The auth schema migration associated with MongoDB 2.6 moved all users into the system.users table in the admin database, but left behind the system.users tables in the other databases (MongoDB 2.6 just ignores these). This seems to cause this assertion when importing into MongoDB 2.6.