mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Joseph Wu <jos...@mesosphere.io>
Subject Re: Review Request 69307: Changed master to hold subscribers in a circular buffer.
Date Wed, 05 Dec 2018 20:53:22 GMT


> On Dec. 4, 2018, 3:52 p.m., Greg Mann wrote:
> > src/master/master.cpp
> > Lines 12109 (patched)
> > <https://reviews.apache.org/r/69307/diff/2/?file=2111388#file2111388line12109>
> >
> >     This seems worthy of LOG(INFO) to me, WDYT?

Sure, I can promote it.  I don't have much of a preference, except that this log line should
_not_ be a warning, since it is fairly easy to hit with a smaller limit set.


> On Dec. 4, 2018, 3:52 p.m., Greg Mann wrote:
> > src/master/master.cpp
> > Lines 12112 (patched)
> > <https://reviews.apache.org/r/69307/diff/2/?file=2111388#file2111388line12112>
> >
> >     Is this ID printed anywhere else? I'm wondering if it's at all useful.

Besides the line written upon subscribing,
```
  LOG(INFO) << "Added subscriber " << http.streamId
            << " to the list of active subscribers";
```
and exiting, 
```
  LOG(INFO) << "Removed subscriber " << id
            << " from the list of active subscribers";
```
the ID is not printed anywhere else.  These log lines were how we spotted the subscriber leak
in the first place.


- Joseph


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/69307/#review211032
-----------------------------------------------------------


On Nov. 30, 2018, 6:49 p.m., Joseph Wu wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/69307/
> -----------------------------------------------------------
> 
> (Updated Nov. 30, 2018, 6:49 p.m.)
> 
> 
> Review request for mesos, Benno Evers, Benjamin Mahler, Gastón Kleiman, and Greg Mann.
> 
> 
> Bugs: MESOS-9258
>     https://issues.apache.org/jira/browse/MESOS-9258
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> This adds a flag (--max_operator_event_stream_subscribers) to the
> master which controls how many active subscribers on the Master's
> event stream will be allowed at any time.
>     
> The default is 1000 subscribers, which is purposefully higher
> than we expect is needed.  Operators aware that their  network
> has clients/proxies whom do not close connections have the
> option of lowering this flag.
> 
> 
> Diffs
> -----
> 
>   docs/configuration/master.md 575476728079d884fe86b1a3a56614647d1b572e 
>   src/master/constants.hpp 76ad0c3b1494cb97888af5545acc998b8fb15bf6 
>   src/master/flags.hpp ed2d76af2831d1b9718724936fd3e4850e99b332 
>   src/master/flags.cpp 2677738d19caa7d600e591aeb4fb37f0c2318d45 
>   src/master/master.hpp c7becfa615964674dcf1ebd9424aa5818a0fdb85 
>   src/master/master.cpp 3b3824a67f46866cd64e32d7f9f92484b5891aa2 
>   src/tests/api_tests.cpp edde48a4b4b46c9f47bc44de389c97b77322b8e8 
> 
> 
> Diff: https://reviews.apache.org/r/69307/diff/2/
> 
> 
> Testing
> -------
> 
> ```
> make check
> 
> src/mesos-tests --gtest_filter="*MaxEventStreamSubscribers*" --gtest_repeat=-1 --gtest_break_on_failure
> ```
> 
> 
> Thanks,
> 
> Joseph Wu
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message