Unix sockets raise exception on long writes #3799

evanphx · 2016-04-12T22:27:13Z

Environment

jruby 9.0.5.0 (2.2.3) 2016-01-26 7bee00d Java HotSpot(TM) 64-Bit Server VM 25.72-b15 on 1.8.0_72-b15 +jit [darwin-x86_64]
Darwin veritas.local 15.4.0 Darwin Kernel Version 15.4.0: Fri Feb 26 22:08:05 PST 2016; root:xnu-3248.40.184~3/RELEASE_X86_64 x86_64
Rails app running puma: https://github.com/cdonadeo/puma_big_response_test

Expected Behavior

Long strings passed to UNIXSocket#write writes the data and returns the number of bytes written.

Actual Behavior

#write writes a subset of the data and raises an exception: #<SystemCallError: Unknown error - >}. The amount of data varies but I've seen 8192 show up not-randomly, almost like there was a 8k buffer somewhere.

The git repo linked to above shows the behavior. The exception in question is raise inside #fast_write. Changing the code to use #write instead of #syswrite does not change the behavior.

The text was updated successfully, but these errors were encountered:

headius · 2016-04-12T22:29:00Z

Nice, haven't had a good UNIXSocket bug in a while. What are UNIX sockets used for in a puma set up?

I've got some long flights today so I may have time to look into this.

TheKidCoder · 2016-04-12T22:30:08Z

Puma can use a UNIX socket to listen/respond to HTTP requests. For example, you can proxy requests from Nginx to Puma over a UNIX socket.

evanphx · 2016-04-12T22:30:59Z

@headius What @TheKidCoder said.

headius · 2016-04-12T22:31:32Z

Ok that makes sense.

ericqweinstein · 2016-04-13T02:47:31Z

Does this issue correspond to this constant?

headius · 2016-04-20T19:28:37Z

@ericqweinstein Seems plausible...I'm not familiar with that value.

headius · 2016-08-22T17:02:23Z

We may be able to make progress on this with @etehtsea helping work on sockets. Sadly it won't be in 9.1.3.0.

etehtsea · 2016-08-31T13:05:32Z

Hopefully, I have a fix for this.

Do the same as in 857b76f

headius · 2016-11-08T06:15:13Z

@etehtsea Got that fix in a releasable state? We're hoping to push 9.1.6.0 tomorrow.

etehtsea · 2016-11-08T10:55:40Z

@headius no, I don't have better solution for now.

headius · 2016-11-08T17:38:27Z

@etehtsea I guess what I was asking is if the commit above is a "good enough" fix. If so, do you want to throw it in a PR for testing?

Status of testing is unclear and this is a key core API. Punting.

etehtsea · 2016-11-09T05:49:12Z

@headius you said earlier that this fix couldn't be "good enough" because possibly it masks real issue.

As far as I remember I tested this issue using this app so if needed I could try to convert it to regression spec.

Fixes jruby#3799

Fix IO#syswrite. Fixes #3799

sumitmah · 2017-01-12T07:02:19Z

@headius I'm still not getting consistent behaviour with 9.1.7.0 for the puma_big_response_test example app created by @evanphx.

sumitmah · 2017-01-13T02:04:47Z

@headius we never merged following commit to master. etehtsea@613c99f

mavenik · 2017-01-30T02:42:21Z

@headius This is still happening on 9.1.7.0 for long writes and Unix sockets with Puma. Switching to TCP port configuration for Puma has resolved it temporarily, but that's not a viable solution IMO.

Any ETA on the fix?

etehtsea · 2017-01-30T12:52:10Z

@mavenik @sumitmah on what OS you caught this?

sumitmah · 2017-01-30T13:31:45Z

@etehtsea I can reproduce on linux. This commit etehtsea@613c99f I've tested, supposed to fix it but it never got merged to master. Only next commit after this was merged.

etehtsea · 2017-01-30T15:03:46Z

@sumitmah @mavenik At some point I decided that I had gone crazy, but actually this issue was fixed after updating to jnr-enxio 0.14 and appeared again after updating to jnr-unixsocket 0.15.
It work properly with jnix-unixsocket 0.14. I'm looking for the fix now.

cc: @enebo @headius

Fixed code from jnr-enxio is not used anymore after updating to jnr-unixsocket 0.15.

See jnr/jnr-unixsocket#38 Fixed code from jnr-enxio is not used anymore after updating to jnr-unixsocket 0.15. Regression spec accidentally passes on TravisCI and Ubuntu xenial. In the meantime there are bug reports that issue still reproduces in some environments. Personally I caught it under macOS.

etehtsea · 2017-01-30T17:03:21Z

@sumitmah @mavenik should be fixed when #4466 will be merged.

Fix #3799 again.

etehtsea · 2017-02-12T15:46:41Z

@sumitmah @mavenik please check again on master

mavenik · 2017-02-13T10:33:55Z

@etehtsea Thanks! Will fetch master and verify this.

evanphx mentioned this issue Apr 12, 2016

Large Responses Truncated Only Via UNIX Socket puma/puma#929

Closed

headius added this to the JRuby 9.1.1.0 milestone Apr 20, 2016

headius modified the milestones: JRuby 9.1.1.0, JRuby 9.1.2.0 May 11, 2016

enebo modified the milestones: JRuby 9.1.2.0, JRuby 9.1.3.0 May 23, 2016

headius modified the milestones: JRuby 9.1.4.0, JRuby 9.1.3.0 Aug 22, 2016

etehtsea added a commit to etehtsea/jruby that referenced this issue Aug 31, 2016

IO#syswrite should always block. Fixes jruby#3799

613c99f

Do the same as in 857b76f

enebo modified the milestones: JRuby 9.1.4.0, JRuby 9.1.5.0, JRuby 9.1.6.0 Sep 1, 2016

headius modified the milestones: JRuby 9.1.7.0, JRuby 9.1.6.0 Nov 8, 2016

etehtsea added a commit to etehtsea/jruby that referenced this issue Nov 16, 2016

Bump jnr-enxio and add regression test

ea714e8

Fixes jruby#3799

headius closed this as completed in bfcec47 Nov 22, 2016

headius added a commit that referenced this issue Nov 22, 2016

Merge pull request #4123 from etehtsea/gh-3799-syswrite-block

723f2da

Fix IO#syswrite. Fixes #3799

etehtsea mentioned this issue Jan 30, 2017

Handle EAGAIN/EWOULDBLOCK in a special way jnr/jnr-unixsocket#38

Merged

etehtsea added a commit to etehtsea/jruby that referenced this issue Jan 30, 2017

Fix jruby#3799 again.

2ff1746

Fixed code from jnr-enxio is not used anymore after updating to jnr-unixsocket 0.15.

headius added a commit that referenced this issue Feb 9, 2017

Merge pull request #4466 from etehtsea/gh-3799-unixsocket

fc8508a

Fix #3799 again.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Unix sockets raise exception on long writes #3799

Unix sockets raise exception on long writes #3799

evanphx commented Apr 12, 2016

headius commented Apr 12, 2016

TheKidCoder commented Apr 12, 2016

evanphx commented Apr 12, 2016

headius commented Apr 12, 2016

ericqweinstein commented Apr 13, 2016

headius commented Apr 20, 2016

headius commented Aug 22, 2016

etehtsea commented Aug 31, 2016

headius commented Nov 8, 2016

etehtsea commented Nov 8, 2016

headius commented Nov 8, 2016

etehtsea commented Nov 9, 2016

sumitmah commented Jan 12, 2017

sumitmah commented Jan 13, 2017

mavenik commented Jan 30, 2017

etehtsea commented Jan 30, 2017

sumitmah commented Jan 30, 2017

etehtsea commented Jan 30, 2017 •

edited

etehtsea commented Jan 30, 2017

etehtsea commented Feb 12, 2017

mavenik commented Feb 13, 2017

Unix sockets raise exception on long writes #3799

Unix sockets raise exception on long writes #3799

Comments

evanphx commented Apr 12, 2016

Environment

Expected Behavior

Actual Behavior

headius commented Apr 12, 2016

TheKidCoder commented Apr 12, 2016

evanphx commented Apr 12, 2016

headius commented Apr 12, 2016

ericqweinstein commented Apr 13, 2016

headius commented Apr 20, 2016

headius commented Aug 22, 2016

etehtsea commented Aug 31, 2016

headius commented Nov 8, 2016

etehtsea commented Nov 8, 2016

headius commented Nov 8, 2016

etehtsea commented Nov 9, 2016

sumitmah commented Jan 12, 2017

sumitmah commented Jan 13, 2017

mavenik commented Jan 30, 2017

etehtsea commented Jan 30, 2017

sumitmah commented Jan 30, 2017

etehtsea commented Jan 30, 2017 • edited

etehtsea commented Jan 30, 2017

etehtsea commented Feb 12, 2017

mavenik commented Feb 13, 2017

etehtsea commented Jan 30, 2017 •

edited