Displaying #maven-dev/2017-03-09.log:

Thu Mar 9 10:01:57 2017  olamy:Joined the channel
Thu Mar 9 11:15:18 2017  Michael-O:Joined the channel
Thu Mar 9 11:57:07 2017  tibor_:Joined the channel
Thu Mar 9 11:58:05 2017  tibor_:Michael-O: Hi Michael. The windcard port was not from Maven main process?
Thu Mar 9 12:21:00 2017  Michael-O:tibor_, no, JVM-internal
Thu Mar 9 12:21:12 2017  Michael-O:I wasn't able to come online yesterday, still owe you the second ponit
Thu Mar 9 13:14:13 2017  tibor_:Joined the channel
Thu Mar 9 16:57:10 2017  Michael-O:Joined the channel
Thu Mar 9 20:41:19 2017  jieryn:Joined the channel
Thu Mar 9 22:58:24 2017  tibor_:Joined the channel
Thu Mar 9 22:58:59 2017  tibor_:Michael-O: Hi Michael
Thu Mar 9 22:59:05 2017  Michael-O:hi
Thu Mar 9 23:00:15 2017  tibor_:Do you have a little time now?
Thu Mar 9 23:00:51 2017  tibor_:I think you fould something.
Thu Mar 9 23:01:02 2017  Michael-O:Tell me, quick. I have 10 min left.
Thu Mar 9 23:01:11 2017  Michael-O:Deep stuff has to wait for tomorrow
Thu Mar 9 23:01:40 2017  tibor_:Tomorrow will be better time. Today my parents visited me so I could not join sooner.
Thu Mar 9 23:02:27 2017  Michael-O:I was busy with my kitchen planning
Thu Mar 9 23:03:08 2017  tibor_:You think we have to make any testing?
Thu Mar 9 23:04:17 2017  Michael-O:Hmm...I think no, unless you know where the race condition could be.
Thu Mar 9 23:04:21 2017  Michael-O:What did you find anyway?
Thu Mar 9 23:04:58 2017  tibor_:I know you mentioned 0.0.0.0 and locks in JVM.
Thu Mar 9 23:05:19 2017  Michael-O:Yes...
Thu Mar 9 23:05:43 2017  tibor_:I thought that static context in a class started Thread and maybe JVM on FreeBSD does not like it .
Thu Mar 9 23:06:21 2017  tibor_:My plan was to start jar file from java app.
Thu Mar 9 23:06:45 2017  Michael-O:Could be something
Thu Mar 9 23:06:51 2017  Michael-O:let's investigate tomorrow
Thu Mar 9 23:07:12 2017  tibor_:Also my plan was to write limitations of the plugin in our doc.
Thu Mar 9 23:07:22 2017  Michael-O:Forking a thread from a static block is weird
Thu Mar 9 23:07:29 2017  Michael-O:what kind of limitations do we have?
Thu Mar 9 23:07:46 2017  tibor_:Long running jvm.
Thu Mar 9 23:08:24 2017  Michael-O:I see
Thu Mar 9 23:08:27 2017  tibor_:VirtualBox with FreeBSD 11 and 9 is fine. Realbox FreeBSD 10 hangs.
Thu Mar 9 23:08:51 2017  tibor_:Is it cause of version of OS, maybe?
Thu Mar 9 23:09:22 2017  tibor_:Is realbox somehow different from virtualization, maybe.
Thu Mar 9 23:09:48 2017  Michael-O:This was my idea too
Thu Mar 9 23:10:03 2017  Michael-O:the real box was previously 9.3 and has been updated
Thu Mar 9 23:10:16 2017  Michael-O:Unfortunately, I have no other real sever with 11 at hand
Thu Mar 9 23:10:16 2017  tibor_:to what version
Thu Mar 9 23:10:32 2017  Michael-O:to 10.3
Thu Mar 9 23:11:41 2017  tibor_:I can run form from parent jvm tomorrow. I guess last test.
Thu Mar 9 23:11:47 2017  tibor_:fork
Thu Mar 9 23:12:11 2017  Michael-O:Yepp
Thu Mar 9 23:12:19 2017  Michael-O:The second issue I have found
Thu Mar 9 23:12:34 2017  tibor_:Maybe not to start daemon Thread from static context, but start from main().
Thu Mar 9 23:12:55 2017  Michael-O:I ran 2.19.2-experimental on Fedora 25
Thu Mar 9 23:13:09 2017  Michael-O:ForkModeIT failed too, just like on FreeBSD
Thu Mar 9 23:13:33 2017  tibor_:I have Fedora too. I would like to try as well.
Thu Mar 9 23:14:08 2017  Michael-O:Then I was reverting every single change of the branch until I have found that OutputStreamFlushableCommandline's flushReceiver was null
Thu Mar 9 23:14:18 2017  Michael-O:this caused failures on Fedora and FreeBSD
Thu Mar 9 23:14:37 2017  Michael-O:removing the comment made it pass on Fedora completely and less failures on FreeBSD
Thu Mar 9 23:15:24 2017  tibor_:I will check it.
Thu Mar 9 23:15:42 2017  tibor_:There is a lot of experiments.
Thu Mar 9 23:16:21 2017  Michael-O:Surprisingly that the missing FlushReceiver causes so many failures
Thu Mar 9 23:17:11 2017  tibor_:it is responsible for flushing the OutputStream which is writing commands to forked jvm
Thu Mar 9 23:18:15 2017  Michael-O:So flush seems to be very nessary
Thu Mar 9 23:18:19 2017  Michael-O:necessary
Thu Mar 9 23:18:26 2017  tibor_:yes
Thu Mar 9 23:18:38 2017  tibor_:it flushes entire ommand
Thu Mar 9 23:18:50 2017  tibor_:not every byte
Thu Mar 9 23:19:05 2017  Michael-O:Linewise is perfectly fine
Thu Mar 9 23:19:09 2017  tibor_:flushing every byte would be expensive
Thu Mar 9 23:19:58 2017  Michael-O:so if no flush is done, the forkedbooter really starves and assumes that the parent VM didn't send someting
Thu Mar 9 23:20:02 2017  Michael-O:so buffers kill us
Thu Mar 9 23:20:03 2017  tibor_:why flushReceiver was null?
Thu Mar 9 23:20:12 2017  Michael-O:Because there was a comment around it
Thu Mar 9 23:20:25 2017  tibor_:aha
Thu Mar 9 23:20:42 2017  Michael-O:line 70
Thu Mar 9 23:23:37 2017  tibor_:and still buggy?
Thu Mar 9 23:24:14 2017  Michael-O:nope, runs flawlessly on my VMs and P4 real box
Thu Mar 9 23:24:20 2017  Michael-O:I will try the box at work tomorrow
Thu Mar 9 23:24:28 2017  tibor_:i mean on fedora
Thu Mar 9 23:24:41 2017  tibor_:i will try master
Thu Mar 9 23:25:05 2017  tibor_:people use Linux, so at least this should be okay
Thu Mar 9 23:25:20 2017  Michael-O:Works flawless on Fedora now
Thu Mar 9 23:25:27 2017  tibor_:aha
Thu Mar 9 23:26:26 2017  tibor_:In my company I use Ubuntu and Jenkins and all is fine
Thu Mar 9 23:26:58 2017  Michael-O:even without flushReceiver?
Thu Mar 9 23:27:14 2017  tibor_:regular release version 2.19.1
Thu Mar 9 23:27:23 2017  tibor_:I did not try snapshot
Thu Mar 9 23:27:35 2017  Michael-O:try the 2.19.2-experimental
Thu Mar 9 23:27:36 2017  tibor_:I can check it out
Thu Mar 9 23:27:53 2017  tibor_:no problem
Thu Mar 9 23:27:57 2017  Michael-O:I can retry on RHEL6 too
Thu Mar 9 23:28:41 2017  tibor_:Did you uncomment Line70?
Thu Mar 9 23:28:47 2017  tibor_:and pushed?
Thu Mar 9 23:28:58 2017  Michael-O:Nope, just locally
Thu Mar 9 23:29:05 2017  Michael-O:please uncomment and push
Thu Mar 9 23:29:09 2017  tibor_:I think it is important
Thu Mar 9 23:29:15 2017  tibor_:I will
Thu Mar 9 23:32:45 2017  tibor_:done
Thu Mar 9 23:33:24 2017  Michael-O:great, will run more tests tomorrow

Comments