[jira] [Commented] (DRILL-3523) Drill client CLI does not reconnect to Drillbit after long period of inactivity
[ https://issues.apache.org/jira/browse/DRILL-3523?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14709637#comment-14709637 ] Daniel Barclay (Drill) commented on DRILL-3523: --- Regarding the seeming timeout and disconnection: I was not able to reproduce that symptom (neither with Drill 1.1 nor with the current state of 1.2, and neither in embedded mode nor local non-embedded mode). If you see the behavior again, please check whether there's anything that looks relevant in the log file (especially when the last visible log activity was, relative to your last user-level interaction), and check what netstat reports about open TCP listener ports and open TCP connections (especially ports 8047 and 31010). Drill client CLI does not reconnect to Drillbit after long period of inactivity --- Key: DRILL-3523 URL: https://issues.apache.org/jira/browse/DRILL-3523 Project: Apache Drill Issue Type: Bug Components: Client - CLI Affects Versions: 1.1.0 Reporter: Hari Sekhon Assignee: Daniel Barclay (Drill) Fix For: 1.2.0 After a period of inactivity in the Drill CLI client before issuing another statement, it appears to time out the session/connection and the client returns an error instead of reconnecting to Drill. It should instead just implicitly !reconnect itself since I can't think of a good reason why I have to do this by hand every time. Unlike in DRILL-3514 the drillbit in this case was never restarted, although the same fix could probably solve both issues. The error received when trying to issue a new query after some absence is: {code}Error: CONNECTION ERROR: Connection /x.x.x.x:54367 -- /x.x.x.x:31010 (user client) closed unexpectedly. [Error Id: 68714cc2-65f5-4cac-b062-44331f8f4c31 ] (state=,code=0) {code} -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (DRILL-3523) Drill client CLI does not reconnect to Drillbit after long period of inactivity
[ https://issues.apache.org/jira/browse/DRILL-3523?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14703247#comment-14703247 ] Hari Sekhon commented on DRILL-3523: More like 30+ mins. I don't know the exact time, as I was working on other things and only noticed this when returning to the drill prompt, after this happened a few times I thought I'd raise it. Since this is drill-embedded there are no intermediate firewalls that could be resetting the connection either. The error that occurs when trying to enter another SQL statement was shown in the original post above. Good point about session alteration. It might require the drill client to remember the session alteration statements that were sent to Drill so it could re-run them on re-connect. This would be a significant usability improvement rather than making the user reconnect and deal with it every time multiple times a day as I was having to do. Drill client CLI does not reconnect to Drillbit after long period of inactivity --- Key: DRILL-3523 URL: https://issues.apache.org/jira/browse/DRILL-3523 Project: Apache Drill Issue Type: Bug Components: Client - CLI Affects Versions: 1.1.0 Reporter: Hari Sekhon Assignee: Daniel Barclay (Drill) Fix For: 1.2.0 After a period of inactivity in the Drill CLI client before issuing another statement, it appears to time out the session/connection and the client returns an error instead of reconnecting to Drill. It should instead just implicitly !reconnect itself since I can't think of a good reason why I have to do this by hand every time. Unlike in DRILL-3514 the drillbit in this case was never restarted, although the same fix could probably solve both issues. The error received when trying to issue a new query after some absence is: {code}Error: CONNECTION ERROR: Connection /x.x.x.x:54367 -- /x.x.x.x:31010 (user client) closed unexpectedly. [Error Id: 68714cc2-65f5-4cac-b062-44331f8f4c31 ] (state=,code=0) {code} -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (DRILL-3523) Drill client CLI does not reconnect to Drillbit after long period of inactivity
[ https://issues.apache.org/jira/browse/DRILL-3523?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14701716#comment-14701716 ] Daniel Barclay (Drill) commented on DRILL-3523: --- After a period of inactivity in the Drill CLI client before issuing another statement, it appears to time out the session/connection After how long a period of inactivity? Is it after around 30 seconds? It don't think the JDBC connection has an intentional timeout, but there's an internal heartbeat/keep-alive check at around 30 seconds that I've seen occur (in my case, when I had the client and server sides configured inconsistently).) and the client returns an error instead of reconnecting to Drill. Which error/exception message? It should instead just implicitly !reconnect itself since I can't think of a good reason why I have to do this by hand every time. It needs to at least not reconnect automatically without telling you, since the server-side state of the connection (e.g., results of USE x, ALTER SESSION ..., etc.) will be reset. (It would be really confusing if the state reset without notice why.) Drill client CLI does not reconnect to Drillbit after long period of inactivity --- Key: DRILL-3523 URL: https://issues.apache.org/jira/browse/DRILL-3523 Project: Apache Drill Issue Type: Bug Components: Client - CLI Affects Versions: 1.1.0 Reporter: Hari Sekhon Assignee: Daniel Barclay (Drill) Fix For: 1.2.0 After a period of inactivity in the Drill CLI client before issuing another statement, it appears to time out the session/connection and the client returns an error instead of reconnecting to Drill. It should instead just implicitly !reconnect itself since I can't think of a good reason why I have to do this by hand every time. Unlike in DRILL-3514 the drillbit in this case was never restarted, although the same fix could probably solve both issues. The error received when trying to issue a new query after some absence is: {code}Error: CONNECTION ERROR: Connection /x.x.x.x:54367 -- /x.x.x.x:31010 (user client) closed unexpectedly. [Error Id: 68714cc2-65f5-4cac-b062-44331f8f4c31 ] (state=,code=0) {code} -- This message was sent by Atlassian JIRA (v6.3.4#6332)