Bitcoin Unlimited v0.12.1bu “Xpedited” -:- Official BU Release

[pekatete]

@Peter Tschipper @sickpig - OK, found the core dump, here's the pastebin.

 

[sickpig]

@Peter Tschipper

wrt stack trace: this kind of pattern happen also on working nodes. on my node it usually happens when I have a peer who is currently not running specified via addnode on the bitcoin.conf file.

@pekatete

How did you generate the output you published on pastebin? I think it's the content of the file you found on /var/crash rather then a real core dump.

At the end of this file it should be the real core dump in a base 64 format. it should start with a line like this:

CoreDump: base64

as Peter said try to look in /var/crash (if it is an ubuntu system)

to confirm if the "kill -11" cmd makes bitcoind segfault launch this command:

dmesg -T | grep bitcoin

if it has crashed for real you should a lime similar ti this one as output:

[Fri Apr 1 16:55:16 2016] bitcoind[13738]: segfault at bd50 ip 000000000000bd50 sp 00007f80621193d8 error 14 in bitcoind (deleted)[400000+228000]

 

[pekatete]

I simply copied the contents from the file in /var/crash .... (aptly named _usr_bin_bitcoind.0.crash)

 

[sickpig]

So it means that there's no core dump at the end of the file. How did you start the bitcoind?

if the file has the core dump at the end this is how it looks around the line I mentioned before:

...
Signal: 6
Uname: Linux 4.4.0-31-generic x86_64
UserGroups: adm cdrom dip lpadmin plugdev sambashare sudo vboxusers
_LogindSession: c2
CoreDump: base64
H4sICAAAAAAC/0NvcmVEdW1wAA==
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
...

now I have to go, though. Will continue later.

 

[pekatete]

This was started at reboot (I have a reboot directive in crontab). As this VPS was the first to crash, I thought the scripts I was paying with caused the crash, so I rebooted it to ensure a clean run. When the other one crashed (as well as this one today), I started from a shell prompt.

PS. Do I need a special editor to see the core dump at the end of the file? I used vi to skim it first, but the pastebin was copied from the default editor in Windows SCP.

 

[pekatete]

Looks like no core dump then .... that's the entire file whichever way I view it.

 

[Peter Tschipper]

@pekatete when you signed up for your VPS did they tell you how many processors you would have?

 

[pekatete]

@Peter Tschipper - They probably did state it somewhere, though can't be sure. They are dirt cheap though, however still have another that I have not updated that is running BU 12.0 (and a webserver) for 15+ weeks without issue.

 

[Peter Tschipper]

@pekatete

can you run the command:

cat /etc/lsb-release

and tell us what the output says exactly...thanks

 

[pekatete]

@Peter Tschipper - Like I said, its a 13 upgraded to a 14 (the pastebin also confirms that). Here's the output from cat/etc/lsb-release:

DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=14.04
DISTRIB_CODENAME=trusty
DISTRIB_DESCRIPTION="Ubuntu 14.04.5 LTS"

 

[Peter Tschipper]

thanks, interesting, our VPS is running on 14.04.4 yours is one up from us...we're still waiting to hear from others.

 

[pekatete]

Again on queue the other VPS crashed. The only data I was able to glean was the strace file (which again has nothing interesting, very similar to the last). Core dump was not logged and nothing out of the ordinary with the debug.log (had not started bitcoind with the empty debug directive, just xthin). I've downgraded this one to 12.0 and restarted. Will wait for the other one to go down again (as it certainly will from what has gone on) and post here the debug.log and core dump (if any this time).

PS. had got used to being prompted to upgrade to 15, here's what I got today (14 now sounds archaic!)

New release '16.04.1 LTS' available
Run 'do-release-upgrade' to upgrade to it.

 

[Peter Tschipper]

thanks for the help....we're running it on our own VPS now, the gitian build, so hopefully we can recreate it in our own env.

 

[pekatete]

Just an update and closure (on my part) on this.
All my VPS's with the updated BU 12.1 crash with regularity and I have been un-able to get the core dumps from following the instructions above on all occasions. If no one has replicated this behaviour, it is probably an issue with my setup.

I'll resort to writing scripts that I run in conjunction with my other regular running tasks to detect when bitcoind crashes / goes to sleep so that they can kill it and restart it.

 

[Peter Tschipper]

@pekatete thanks for all your doing...but for now you can just go back an run 12.0. We were able to get a backtrace from one of our nodes and have a fix which we're testing. Hopefully we'll have bug fix release soon. Thanks again.

 

[pekatete]

@Peter Tschipper - no worries .... at least I now know it is not down to my setup. Since they are not mining nodes, I'll just leave them running 12.1 if I can get the recovery scripts going. And by the way, I can spare one VPS to test your fix .......

 

[HelloGuy]

@pekatete thanks for all your doing...but for now you can just go back an run 12.0. We were able to get a backtrace from one of our nodes and have a fix which we're testing. Hopefully we'll have bug fix release soon. Thanks again.

Great. I am looking forward to your new release and run it!

 

[Christoph Bergmann]

@79b79aa8 We do need some more doc...one relatively quick and easy thing to get the ball rolling would be an install guide...similar to what Classic has but you could also add some of the tricky issues like install on Ubuntu, @sickpig might have some insight there. Do you have access to our slack channels? that's where we hang out mostly.

Yes, more docs would be a good thing. Also you should think about updating the website. For example it lacks information about persons in the team of unlimited. And specifications ...

If you want, I can help with writing docs about features. I'll need technical input, but we can try to write documentations (about unlimited's approach to blocksize, and xthing etc.). But I'm no coder, so I can't express it on the side of the code, just logical abstractions.

 

[Peter Tschipper]

@Christoph Bergmann Thanks for the offer. We've already started updating our docs...we have the xpedited and xthin "getting the most out of them" setup docs...as well as the xthin detailed protocol doc They are in the /doc folder of the source code but I think we should have links to them on our website. maybe you could help there? Also we have doc regarding the setting up of the dev env on Native Windows for anybody who wants to do development..that could also be on our website...of course the BUIP's and team list... You might want to join our slack channel if you haven't done so by asking @theZerg ...it would be easier to communicate there....anything else you can think of is most welcome, even if it's just an idea.

 

[Christoph Bergmann]

Hm, yes, how can I join the slack channel?

Do you have a link to your docs? Maybe I can help editing ...