GitHub user xiaosuo opened a pull request:
https://github.com/apache/thrift/pull/1239
Fix wrong numActiveProcessors_ when failing to notify any IO thread
We should decrease numActiveProcessors_ when failing to notify any
IO thread.
You can merge this pull request into a Git
Github user xiaosuo commented on the issue:
https://github.com/apache/thrift/pull/1211
I think this patch doesn't introduce any inconsistency. And the correctness
of the number of active processors deserves another separated patch. We'd
better not mix fixes into a huge patch
Github user xiaosuo commented on the issue:
https://github.com/apache/thrift/pull/1211
@jeking3 Did you mean that we should fix all issues about the number of
active processors in this patch instead of a separated one?
---
If your project is set up for it, you can reply
Github user xiaosuo commented on the issue:
https://github.com/apache/thrift/pull/1216
However when closing the connection, we don't check if eventFlags_ is zero.
For a new connection, event_ isn't initialized by the constructor() or
init(), and if we fail to pass
Github user xiaosuo commented on the issue:
https://github.com/apache/thrift/pull/1217
I checked other call sites, and found no issue. Because we will close the
connection in the destructor, we won't leak fds at least.
---
If your project is set up for it, you can reply
GitHub user xiaosuo opened a pull request:
https://github.com/apache/thrift/pull/1217
TSocket: Don't close connections when failing to read/write
Sockets maybe registered to a event base of libevent, so the
following epoll_ctl(2) may fail due to EBADF, or may
delete other
GitHub user xiaosuo opened a pull request:
https://github.com/apache/thrift/pull/1216
TNonblockingServer: Fix using uninitialized event_
When there are more than one IO threads, and we have failed to notify
one IO threads, then we have to close the connection. But the event_
Github user xiaosuo commented on the issue:
https://github.com/apache/thrift/pull/1211
@jeking3 you are correct. In fact, we had found this issue, and we should
do such things at all the error path when notifying the io threads. I will
address this issue in a separated patch
Github user xiaosuo commented on the issue:
https://github.com/apache/thrift/pull/1211
I had tried to find my password, but nothing was received. It seems that
Jira doesn't work properly with gmail. Could you help me?
---
If your project is set up for it, you can reply to this email
GitHub user xiaosuo opened a pull request:
https://github.com/apache/thrift/pull/1211
Fix use closed(freed) connections
When failing to add tasks into the thread manager, we close the
corresponding connections, then set the flags of these connections,
which have been already
GitHub user xiaosuo opened a pull request:
https://github.com/apache/thrift/pull/1210
Fix fd leak when failing to dispatch new connections
When failing to dispatch new connections to other IO threads other
than the number 0, we returned these connections for reuse without
Github user xiaosuo commented on the issue:
https://github.com/apache/thrift/pull/422
This patch may introduce a more serious problem in theory.
No standard defines writing to a socket is an atomic operation, so if
writing is interleaved by other threads, the reading end
12 matches
Mail list logo