#weavenetwork

ah it feels so close. When I ping a weave network address from host A and do a tcpdump on host B I think I can see the ping responses coming back... but I guess they're not getting routed appropriately back to host B. http://pastebin.com/mwvpW7hR

hi there

I'm trying to run weave on an ubuntu VM which is running on Azure

I get a died container back

any ideas?

andreaturli: what does `docker logs weave` say?

hi squaremo, sorry I killed that machine. Will ping you as soon as it is up again, ok?

ok. In the meantime it is worth checking you have the latest weave script

errordeveloper: hi there

strange behaviour with zookeeper: it's unable to resolve weave-dns names correctly

nsswitch.conf: hosts: dns files

when I docker exec to the zookeeper host, and ping the other hosts with weave dns - it works..

so it must be a Java thing

I swear it worked until last week. have no idea what changed

oh cr*p the answer is right here... nsswitch should have only dns, right?

ah no. it should work..

yaronr: Is it (related to) the thing where Java looks up the hostname to get an IP address maybe?

squaremo: yes, but the nsswitch hack should have taken care of it.

hi squaremo

I'm using the latest weave script

and after `sudo weave launch` I get `Container 3e30e728d5b1e3b964b2e33635597fbc0166b21d51e655c89db6a3a77243b06d died`

and `docker logs weave` shows

squaremo: sorry I was disconnected

andreaturli: please post the output of `sh -x weave launch`

with a little more imagination please :)

sh -x `which weave` launch

ops sorry

and a little more

sudo sh -x `which weave` launch

;)

is errordeveloper here?

yaronr: not sure I follow

andreaturli: Not related to you. There's a user called errordeveloper, and I'm looking for him

oh ok

moonfish1: interstingly if I run `docker run --privileged -d --name=weave -p 6783:6783/tcp -p 6783:6783/udp zettio/weave:0.8.0 -name 7a:72:95:32:64:6a -iface ethwe` manually, it works

problem found.

run `sudo weave setup`.

and then try launch again.

also, you may want to run `weave reset` first.

reset, setup and then launch?

same error

hmm.

after `sudo weave setup` I can see with `ipconfig`

is it ok?

weave setup is meant to download some images. did it do anything?

oh yes

ok. but weave reset + launch still fails?

time to modify the weave script... Find https://github.com/zettio/weave/blob/v0.8.0/wea... and insert

cat $IP_TMPERR >&2

then run `sudo weave launch` again.

(insert == insert after that line)

?

yaronr: hi

andreaturil: hello?

yaronr: did you look in weavedns logs?

ok moonfish1

trying to edit

seems that I have a different version of the script

how so?

dunno but line 198 is not the same

anyway the extra line prints `RTNETLINK answers: No such process`

how long does it take before it prints that? is it instant or does it take a few secs?

It is highly disturbing that you have a different script. How did you download it? and would you mind uploading it somewhere so I can figure out how it differs?

I followed the https://github.com/zettio/weave/blob/master/REA...

it takes few secs to print that

interesting.

pls upload it the script.

that script is identical to what I linked to; line 198 is exactly as expected.

sorry I may have misread

anyway back to the `RTNETLINK answers: No such process`

quite :)

what does it mean if it is instant or take a few secs?

please add an

cat $IP_TMPOUT

after the previous line

actually, don't do that.

instead run it with -x as before,

i.e. `sudo sh -x `which weave` launch`

I can only add that it remains on L107 for few secs, if this could help

right

I suspect the execution ethtool inside the weavetools container is taking a long time. but I cannot see why that would be the case.

what does `sudo time docker run --privileged --net=host zettio/weavetools:0.8.0 /bin/ethtool -K eth0 tx on` return?

$ ethtool --version

ethtool version 3.13

is that the expected version?

doesn't matter anymore. as per the 0.8.0 release notes. we run our own packaged version now.

hence the invocation via a container.

pls run the above `time docker ...` command.

sorry I missed that, sure

$ sudo time docker run --privileged --net=host zettio/weavetools:0.8.0 /bin/ethtool -K eth0 tx on

0.03user 0.00system 0:02.46elapsed 1%CPU (0avgtext+0avgdata 11908maxresident)k

0inputs+0outputs (0major+844minor)pagefaults 0swaps

moonfish1: ^^

ouch.

that is 2.46 seconds, right?

yes

are you running on some really slow VM?

possibly, it's a 2 core with 3 GB ram