[SERVER-6872] scoped connection not being return to the pool Created: 27/Aug/12 Updated: 15/Feb/13 Resolved: 02/Sep/12 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | None |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Major - P3 |
| Reporter: | Anton V. Volokhov | Assignee: | Jeremy Mikola |
| Resolution: | Duplicate | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Environment: |
mongodb 2.0.6 |
||
| Issue Links: |
|
||||||||
| Operating System: | Linux | ||||||||
| Participants: | |||||||||
| Description |
|
mongodb configuration: 2shards, 3instances each, 3 config servers, remote mongos, 52mbit network between mongos and mongod, 1gbit between mongod's I'm trying to retrieve data from mongodb via single .find() query in mongo console. ;print while attempts to do so on the local machine are always successful, the query on remote mongos fails repeatable on the same entity. console output: db.load.count(): db.stats(): mongos> db.stats(1024*1024); , }, mongos logs: Mon Aug 27 18:49:16 [conn1] want cursor : 1975918310978336011 |
| Comments |
| Comment by Anton V. Volokhov [ 01/Sep/12 ] | |||
|
they say, it is related to https://jira.mongodb.org/browse/SERVER-5157 | |||
| Comment by Anton V. Volokhov [ 29/Aug/12 ] | |||
|
output from mongos during the last run: 0x23433e0 Tue Aug 28 20:06:20 [conn3] have to set shard version for conn: 0x2344d20 ns:test.load my last seq: 0 current: 2 version: 86|1 manager: 0x7f0694004c90 0x2344d20 Tue Aug 28 20:06:20 [conn3] slave '' is not initialized or invalid | |||
| Comment by Anton V. Volokhov [ 28/Aug/12 ] | |||
|
no, the process wasn't stopped. some more info for previous run: this run timing is even more strange: the rest of log is equal to previous. | |||
| Comment by Scott Hernandez (Inactive) [ 28/Aug/12 ] | |||
|
Can you post the server logs from that run?
These events are very far apart; are you sure that you are not stopping somewhere during your cursor next() loop? | |||
| Comment by Anton V. Volokhov [ 28/Aug/12 ] | |||
|
The same behaviour with your query. And same errors on mongod's | |||
| Comment by Scott Hernandez (Inactive) [ 28/Aug/12 ] | |||
|
Please test that with the batchSize; it will return batches from the cursor with 3 docs at a time. | |||
| Comment by Anton V. Volokhov [ 28/Aug/12 ] | |||
|
logs on load1ft Mon Aug 27 18:18:36 [clientcursormon] killing old cursor 6099439472920059228 test.load idle:604000ms Is there any way to increase cursor timeout, or we have to split the data and retrieve it by pieces? all instances in rset01 and primary in rset02 got following error in time interval 18:49:15 - 18:49:20 | |||
| Comment by Scott Hernandez (Inactive) [ 28/Aug/12 ] | |||
|
Are there any messages on the mongod (shards) at this time? This sounds like a cursor timeout (they timeout after 10 minutes of inactivity) in processing the documents in the loop, and that it is related to the network/latency.
| |||
| Comment by Anton V. Volokhov [ 27/Aug/12 ] | |||
|
well, almost the same entity. +- 1000 in 10 runs. |