[MXS-587] readwritesplit module - intensive load - SQLSTATE[HY000]: General error: 2013 Lost connection to MySQL server during query Created: 2016-02-23 Updated: 2016-05-31 Resolved: 2016-05-31 |
|
| Status: | Closed |
| Project: | MariaDB MaxScale |
| Component/s: | N/A |
| Affects Version/s: | 1.2.1, 1.3.0 |
| Fix Version/s: | N/A |
| Type: | Bug | Priority: | Major |
| Reporter: | Mithun | Assignee: | Timofey Turenko |
| Resolution: | Incomplete | Votes: | 0 |
| Labels: | None | ||
| Environment: |
Ubuntu 14.04 x64_64 LTS + MariaDB Galera + Maxscale |
||
| Attachments: |
|
||||||||
| Issue Links: |
|
||||||||
| Description |
|
The module readwritesplit works perfectly for small queries. An operation of reading 80,000 rows from the db and processing it and writing to the db causes maxscale to return: SQLSTATE[HY000]: General error: 2013 Lost connection to MySQL server during query Meanwhile if readconnroute module is used for the above purpose. It does it without any errors. EDIT1 [Splitter Service] [Splitter Listener] [devdb01] [Galera Monitor] [CLI] [CLI Listener] Log |
| Comments |
| Comment by Dipti Joshi (Inactive) [ 2016-02-23 ] | ||
|
johan.wikman, markus makela have we looked at this ? | ||
| Comment by markus makela [ 2016-02-23 ] | ||
|
Please provide logs and the configuration file you used when this happened. Usually there's a reason and that reason is in the error logs in /var/log/maxscale/. If nothing suspicious is in there, this would seems to be a valid bug. | ||
| Comment by Mithun [ 2016-02-24 ] | ||
|
I have updated the logs and the config used. Thanks. | ||
| Comment by markus makela [ 2016-02-24 ] | ||
|
The log has no errors in it so I assume none were logged. This could be related to | ||
| Comment by markus makela [ 2016-04-15 ] | ||
|
tuxmithun Can you try this with the 1.4.1 release of MaxScale? I think that this could be caused by a network error between MaxScale and a slave. When a SELECT query is sent to a slave and the connection to that slave is lost before the result returns, the client is disconnected. If you could try to confirm that a network error does occur between MaxScale and one of the slaves, I think we could close this as Not A Bug. | ||
| Comment by Xavier [ 2016-04-25 ] | ||
|
I have exactly the same problem. All in Docker. Slave DB and MaxScale in same container. Can reproduce the error anytime, siege the webpage and you will see 503. In my case, magento.. clearly returns the same error. I had produced same tests with the normal routing, and does not produce such errors. – Tested version 1.3.0 Gonna update and test with 1.4.1 | ||
| Comment by Mithun [ 2016-05-24 ] | ||
|
markus makela I have tested with maxscale version 1.4.3 with the following error:
This is inspite of the following permission:
| ||
| Comment by markus makela [ 2016-05-24 ] | ||
|
Do you have anything in MaxScale's error logs? | ||
| Comment by Mithun [ 2016-05-24 ] | ||
|
| ||
| Comment by markus makela [ 2016-05-24 ] | ||
|
Can you try to add the following grant to the MaxScale user:
If that's the only error and you are able to sometimes use it, this is probably unrelated. | ||
| Comment by Mithun [ 2016-05-24 ] | ||
|
markus makela done that, its working as of now. Will update here if any more issues. | ||
| Comment by Johan Wikman [ 2016-05-31 ] | ||
|
Problem seems to have been solved. If re-occurs, please create a new issue. |