[
https://issues.apache.org/jira/browse/IMPALA-12511?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Quanlong Huang updated IMPALA-12511:
------------------------------------
Description:
When the client has no activities but keep the connection opened, coordinator
will log the following message every 30 seconds:
{noformat}
I1017 17:14:55.800310 195900 thrift-util.cc:106] TSocket::read() THRIFT_EAGAIN
(timed out) after %f ms: Unknown error 30000
I1017 17:15:25.830340 195900 thrift-util.cc:106] TSocket::read() THRIFT_EAGAIN
(timed out) after %f ms: Unknown error 30000
{noformat}
30000 here is the timeout value but not an "Unknown error". The confusing log
is fixed in THRIFT-5670. We should backport it to our native-toolchain.
With the fix, the logs will become:
{noformat}
I1023 08:47:20.240015 25785 thrift-util.cc:198] TSocket::read() THRIFT_EAGAIN
(timed out) after 30000 ms
I1023 08:47:50.269630 25785 thrift-util.cc:198] TSocket::read() THRIFT_EAGAIN
(timed out) after 30000 ms
{noformat}
Users can find idle clients based on these logs.
was:
When the client has no activities but keep the connection opened, coordinator
will log the following message every 30 seconds:
{noformat}
I1017 17:14:55.800310 195900 thrift-util.cc:106] TSocket::read() THRIFT_EAGAIN
(timed out) after %f ms: Unknown error 30000
I1017 17:15:25.830340 195900 thrift-util.cc:106] TSocket::read() THRIFT_EAGAIN
(timed out) after %f ms: Unknown error 30000
{noformat}
30000 here is the timeout value but not an "Unknown error". The confusing log
is fixed in THRIFT-5670. We should backport it to our native-toolchain.
> Confusing timeout logs due to THRIFT-5670
> -----------------------------------------
>
> Key: IMPALA-12511
> URL: https://issues.apache.org/jira/browse/IMPALA-12511
> Project: IMPALA
> Issue Type: Bug
> Reporter: Quanlong Huang
> Assignee: Quanlong Huang
> Priority: Major
>
> When the client has no activities but keep the connection opened, coordinator
> will log the following message every 30 seconds:
> {noformat}
> I1017 17:14:55.800310 195900 thrift-util.cc:106] TSocket::read()
> THRIFT_EAGAIN (timed out) after %f ms: Unknown error 30000
> I1017 17:15:25.830340 195900 thrift-util.cc:106] TSocket::read()
> THRIFT_EAGAIN (timed out) after %f ms: Unknown error 30000
> {noformat}
> 30000 here is the timeout value but not an "Unknown error". The confusing log
> is fixed in THRIFT-5670. We should backport it to our native-toolchain.
> With the fix, the logs will become:
> {noformat}
> I1023 08:47:20.240015 25785 thrift-util.cc:198] TSocket::read() THRIFT_EAGAIN
> (timed out) after 30000 ms
> I1023 08:47:50.269630 25785 thrift-util.cc:198] TSocket::read() THRIFT_EAGAIN
> (timed out) after 30000 ms
> {noformat}
> Users can find idle clients based on these logs.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
