Yep you found it. You were unlucky enough to install a version of master that had this regression. How did you install requests? Using pip should also install the necessary dependencies. Either way everything should still work, you'll just have to manually start/stop indexing since you won't ever be able to stop it.
You are correct about HDP 2.6. On Tue, Feb 27, 2018 at 12:57 PM, David McGinnis < mcginn...@avalonconsult.com> wrote: > Looks like this issue was introduced in this commit: > https://github.com/apache/metron/commit/1c9437c41c4eb626bd44 > 193ec6545f9356f3f7b2 . Looking past this, there is only one commit that > seems to deal with this, but it is for the Ansible deployment, so I don't > believe it would affect an MPack installation: > https://github.com/apache/metron/commit/fa86663ecdb166f889c9 > c0c4c34373aec68dbd30 . > > I installed a newer version of requests just to see if that would fix the > issue, and I'm getting a similar issue in a different package relied upon > by requests now (urllib3). It seems as if something is causing sys to not > exist or not point to what it should. I wonder if it has to do with the > python2.6 vs python2.7 issue. > > Also, as I was looking through the commits, it looks like using the last > official release wouldn't work for me, since we are on HDP 2.6, and changes > went in since the last release to support that version of HDP. > > On Tue, Feb 27, 2018 at 12:12 PM, David McGinnis < > mcginn...@avalonconsult.com> wrote: > >> *Simon:* >> >> There are some errors from processing showing up in the indexing topic, >> but no data is making it through without restarting Metron completely >> (restarting just the indexing service doesn't seem to do it). >> >> I don't have easy access to the Storm UI at the moment. I have a work >> around to get the logs that I'm going to look at shortly. >> >> I understand, but I didn't realize I had not pulled from a release until >> I had already installed, and this is just a prototype cluster anyways. I'll >> likely pull from an established release for the production deployment. >> >> *Ryan:* >> >> I actually found something interesting in the ambari-agent logs on the >> Metron node. See the callstack below. >> >> Traceback (most recent call last): >> File >> "/usr/lib/python2.6/site-packages/ambari_agent/PythonReflectiveExecutor.py", >> line 59, in run_file >> imp.load_source('__main__', script) >> File "/var/lib/ambari-agent/cache/common-services/METRON/0.4.3/pa >> ckage/scripts/indexing_master.py", line 18, in <module> >> import requests >> File "/usr/lib/python2.7/site-packages/requests/__init__.py", line 53, >> in <module> >> from .packages.urllib3.contrib import pyopenssl >> File "/usr/lib/python2.7/site-packages/requests/packages/__init__.py", >> line 61, in load_module >> if name in sys.modules: >> AttributeError: 'NoneType' object has no attribute 'modules' >> >> Perhaps there is a missing component that needs to be installed? >> >> On Tue, Feb 27, 2018 at 12:01 PM, Ryan Merriman <merrim...@gmail.com> >> wrote: >> >>> If there was an error happening during the service check, it will be in >>> the Ambari agent logs. I would look there first. >>> >>> On Feb 27, 2018, at 11:45 AM, Simon Elliston Ball < >>> si...@simonellistonball.com> wrote: >>> >>> Anything appearing on the indexing topic in kafka? >>> >>> Anything in the logs for the indexing topology in Storm UI? >>> >>> Master isn’t always the best place to start, might be worth sticking to >>> a released build to kick the tyres. >>> >>> Simon >>> >>> On 27 Feb 2018, at 17:38, David McGinnis <mcginn...@avalonconsult.com> >>> wrote: >>> >>> All, >>> >>> I have a kerberized HDP 2.6.3 cluster which I have installed Metron on >>> through an MPack. I am using code straight from the master branch of the >>> github as of last week. >>> >>> When I start Metron, all of the components correct start up. Shortly >>> after, however, the Indexing Service stops working. I have not been able to >>> find any logs that seem to capture any issues, and both indexing storm >>> topologies are still active and running after the failure. Those topologies >>> only report a warning about one of my parsers using the default HDFS writer >>> config, which appears to be normal. >>> >>> Not only is this making my dashboard red when it shouldn't be, but also >>> it seems as if data doesn't go through the pipeline until the entire Metron >>> system is restarted. I am not confident this is related, but I suspect it >>> is. >>> >>> Where should I go to find logs about this service, or does anyone have >>> any ideas why this would be happening? >>> >>> -- >>> David McGinnis >>> Staff Hadoop Consultant | Avalon Consulting, LLC >>> <http://www.avalonconsult.com/>M: (513) 439-0082 >>> LinkedIn <http://www.linkedin.com/company/avalon-consulting-llc> | >>> Google+ <http://www.google.com/+AvalonConsultingLLC> | Twitter >>> <https://twitter.com/avalonconsult> >>> ------------------------------------------------------------ >>> ------------------------------------------------- >>> This message (including any attachments) contains confidential >>> information >>> intended for a specific individual and purpose, and is protected by law. >>> If >>> you are not the intended recipient, you should delete this message. Any >>> disclosure, copying, or distribution of this message, or the taking of >>> any >>> action based on it, is strictly prohibited. >>> >>> >>> >> >> >> -- >> David McGinnis >> Staff Hadoop Consultant | Avalon Consulting, LLC >> <http://www.avalonconsult.com/>M: (513) 439-0082 >> LinkedIn <http://www.linkedin.com/company/avalon-consulting-llc> | >> Google+ <http://www.google.com/+AvalonConsultingLLC> | Twitter >> <https://twitter.com/avalonconsult> >> ------------------------------------------------------------ >> ------------------------------------------------- >> This message (including any attachments) contains confidential >> information >> intended for a specific individual and purpose, and is protected by law. >> If >> you are not the intended recipient, you should delete this message. Any >> disclosure, copying, or distribution of this message, or the taking of >> any >> action based on it, is strictly prohibited. >> > > > > -- > David McGinnis > Staff Hadoop Consultant | Avalon Consulting, LLC > <http://www.avalonconsult.com/>M: (513) 439-0082 > LinkedIn <http://www.linkedin.com/company/avalon-consulting-llc> | Google+ > <http://www.google.com/+AvalonConsultingLLC> | Twitter > <https://twitter.com/avalonconsult> > ------------------------------------------------------------ > ------------------------------------------------- > This message (including any attachments) contains confidential information > intended for a specific individual and purpose, and is protected by law. > If > you are not the intended recipient, you should delete this message. Any > disclosure, copying, or distribution of this message, or the taking of any > action based on it, is strictly prohibited. >
