steveloughran commented on PR #4766:
URL: https://github.com/apache/hadoop/pull/4766#issuecomment-1232730741
ok, i'm merging this; mukund's vote, mehakmeets comments and yetus are all
happy
--
This is an automated message from the Apache Git Service.
To respond to the message, please log
steveloughran commented on PR #4766:
URL: https://github.com/apache/hadoop/pull/4766#issuecomment-1228433169
tested, s3 london
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific
steveloughran commented on PR #4766:
URL: https://github.com/apache/hadoop/pull/4766#issuecomment-1227090021
need some reviews here. this is a critical bug
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above
steveloughran commented on PR #4766:
URL: https://github.com/apache/hadoop/pull/4766#issuecomment-1225489007
either the fs tracks all open streams (weak ref map) and scans them, or each
stream schedules a worker to run every few minutes which will release the
stream if idle. fs is probably
steveloughran commented on PR #4766:
URL: https://github.com/apache/hadoop/pull/4766#issuecomment-1223780723
if there already is a field there, yes. i've been wondering if there was a
way to do some retirement of long-lived input streams. it doesn't matter so
much with vectored io or
steveloughran commented on PR #4766:
URL: https://github.com/apache/hadoop/pull/4766#issuecomment-1222517838
the last patch factors out stream draining; adds test for corner cases,
especially escalation from read to abort. used in classic and prefetching
streams, with unit tests.
steveloughran commented on PR #4766:
URL: https://github.com/apache/hadoop/pull/4766#issuecomment-1221329260
as this draining code is used in prefetch too, i'm going to
1. create a StreamDrainer class which the prefetch stream will also switch
to; this code is fussy and I don't want
steveloughran commented on PR #4766:
URL: https://github.com/apache/hadoop/pull/4766#issuecomment-1221019821
tested: s3 london
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific
steveloughran commented on PR #4766:
URL: https://github.com/apache/hadoop/pull/4766#issuecomment-1221019682
HADOOP-18410. copy fields to variables before use in a lambda expression
the key cause is that even though the fields passed in to drain() were
converted to references
steveloughran commented on PR #4766:
URL: https://github.com/apache/hadoop/pull/4766#issuecomment-1220968467
The async stream draining only executes (in the other thread) if the
original invoker waits for the result.
this is not caused by some synchronized conflict -I made
10 matches
Mail list logo
