0

I have installed mongodb 3.6 standalone instance and created some collections through insert query in mongo shell. My app points to the db server, the app creates some new collections, it reads, writes and so on. This is running for more than two months. If we restart the instance we can see in mongodb collections which are created through insert query are lost. Now we can see the same behavior in other environments. Also, while analyzing this problem, we found that during mongodb server start there was a error saying "mongodb failed to create WiredTiger bulk cursor: Device or resource busy". Any idea what that error line implies?

The below messges are found in mongodb log file during server shutdown

**2018-08-28T12:44:46.138+0530 I CONTROL [signalProcessingThread] got signal 1 (Hangup), will terminate after current cmd ends**

**2018-08-28T12:44:46.138+0530 I NETWORK [signalProcessingThread] shutdown: going to close listening sockets...**

**2018-08-28T12:44:46.138+0530 I NETWORK [signalProcessingThread] removing socket file: /tmp/mongodb-27017.sock**

**2018-08-28T12:44:46.152+0530 I FTDC [signalProcessingThread] Shutting down full-time diagnostic data capture**

**2018-08-28T12:44:46.158+0530 I STORAGE [signalProcessingThread] WiredTigerKVEngine shutting down**

**2018-08-28T12:44:46.265+0530 I STORAGE [signalProcessingThread] shutdown: removing fs lock...**

**2018-08-28T12:44:46.265+0530 I CONTROL [signalProcessingThread] now exiting**

**2018-08-28T12:44:46.266+0530 I CONTROL [signalProcessingThread] shutting down with code:0**

And during server startup the below messages are found in the log file,

**2018-08-28T14:19:39.256+0530 I STORAGE [thread2] createCollection: config.system.sessions with generated UUID: e981a966-4f87-45a9-9adb-6b7e2fdb8803**

**2018-08-28T14:19:39.277+0530 I INDEX [thread2] build index on: config.system.sessions properties: { v: 2, key: { lastUse: 1 }, name: "lsidTTLIndex", ns: "config.system.sessions", expireAfterSeconds: 1800 }**

**2018-08-28T14:19:39.277+0530 I INDEX [thread2] building index using bulk method; build may temporarily use up to 500 megabytes of RAM**

**2018-08-28T14:19:39.278+0530 W STORAGE [thread2] failed to create WiredTiger bulk cursor: Device or resource busy**

**2018-08-28T14:19:39.278+0530 W STORAGE [thread2] falling back to non-bulk cursor for index table:index-12--7471022903774062594**

**2018-08-28T14:19:39.278+0530 I INDEX [thread2] build index done. scanned 0 total records. 0 secs**
Improve this question
5
  • 1
    The error message implies an I/O issue, but you'll need to share more details. What O/S version and filesystem are you using, and how is your dbPath mounted? Are you shutting down mongod before restarting the instance? Also, what specific version of MongoDB server is this (as reported by db.version() in the mongo shell)? Lastly, are there any messages in the log file for your mongod server during the shutdown or startup sequence?
    – Stennie
    Commented Sep 17, 2018 at 5:31
  • Hi Stennie, Thanks for your response. There are some messages during server shutdown and startup. I have edited the question by including it. Please find the same.
    – Rilwan
    Commented Sep 17, 2018 at 6:26
  • 1
    Thanks for adding some shutdown log lines; these appear to be normal. Can you add the additional information I suggested: What O/S version and filesystem are you using, and how is your dbPath mounted? What specific version of MongoDB server is this (as reported by db.version() in the mongo shell)? The message during server startup does not suggest any missing collections. It appears there was a transient error building the index on config.system.sessions but the following log lines show that index build retrying and completing.
    – Stennie
    Commented Sep 17, 2018 at 6:57
  • Hi Stennie, I have simulated the issue. This is because of the application
    – Rilwan
    Commented Sep 17, 2018 at 13:04
  • @Rilwan, What is MongoDB version(x,y,z)?
    – Md Haidar Ali Khan
    Commented Sep 18, 2018 at 5:06

1 Answer 1

Reset to default
-1

I have gone through your error log and find out the MongoDB Bug affected version error , which are able to see in your error log such as:

"mongodb failed to create WiredTiger bulk cursor: Device or resource busy"

As per MongoDB jira documentation here as it is Server Bug having affected version was 2.8.0-rc0.

The Bug fix version of MongoDB is 2.8.0-rc1.

And the second Bug error, which is

2018-08-28T12:44:46.138+0530 I CONTROL [signalProcessingThread] got signal 1 (Hangup), will terminate after current cmd ends

As jira documentation here It's affected Bug in MongoDB version 3.0.12.

As per @RamonFernandez from jira documentation, seems like mongod is receiving the SIGHUP signal and terminating, which is expected behavior. You need to check who or what is sending this signal.

Improve this answer
3
  • 1
    Jira is an issue tracker, not documentation. Similar issues reported by other users are not necessarily confirmed bugs or even the same problem. The first issue you noted (SERVER-16236) was specific to a very early release candidate for MongoDB 3.0. The device busy message is similar because it comes from the operating system, but in the SERVER report this was part of a fatal error while in the DBA question it was a warning (the index build completed). The second issue (SERVER-24282) was resolved as "works as designed" because it is not a bug (a hangup signal was sent to the mongod process).
    – Stennie
    Commented Sep 23, 2018 at 21:53
  • @Stennie,Thanks for your advice. May be I will be right or wrong , As know that the documentation is something related to collection of particular subject. As Jira is definitely is Issue Tracker site blog. But if somebody will give that jira web reference,then what should be his/her reference point? It will be "Jira documentation" or "Jira Tracking Documentation" or else?
    – Md Haidar Ali Khan
    Commented Sep 24, 2018 at 10:52
  • 1
    For clarity I'd just call it the "MongoDB issue tracker", since Jira isn't the MongoDB documentation or the MongoDB blog. Documentation and blogs are curated content, with documentation focused on reference information and blog posts more topic-based discussion. Issues in Jira are a mix of bugs, features, tasks, and discussion. Some careful interpretation is often required to distinguish the scope and relevance of Jira issues and separate fact from opinion or speculation.
    – Stennie
    Commented Sep 24, 2018 at 13:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.