| *** rfolco has joined #softwarefactory | 00:21 | |
| *** rfolco has quit IRC | 01:03 | |
| sfbender | Yatin Karel created DLRN master: Use git-core instead of git https://softwarefactory-project.io/r/20053 | 07:51 |
|---|---|---|
| *** jpena|off is now known as jpena | 08:09 | |
| sfbender | Merged DLRN master: Use git-core instead of git https://softwarefactory-project.io/r/20053 | 08:28 |
| *** apevec has joined #softwarefactory | 09:07 | |
| sfbender | Merged www.softwarefactory-project.io master: Add sprint 15 summary https://softwarefactory-project.io/r/20046 | 09:49 |
| lyr | How incompatible is 3.4 with CentOS 8 ? | 10:20 |
| lyr | Ansible's asking dnf instead of yum for python 3, python3-dnf is only available on CentOS 8... I'm in a deadlock on CentOS 7 atm | 10:21 |
| *** zenkuro has joined #softwarefactory | 10:50 | |
| *** zenkuro has quit IRC | 11:27 | |
| *** zenkuro has joined #softwarefactory | 11:28 | |
| *** rfolco has joined #softwarefactory | 11:39 | |
| *** zenkuro has quit IRC | 11:53 | |
| *** zenkuro has joined #softwarefactory | 11:53 | |
| *** jpena is now known as jpena|lunch | 12:32 | |
| tristanC | zenkuro: lyr: sf is packaged for centos7, support for centos8 or fedora is still under discussion, i think the main blocker is to provide a simple upgrade path for existing deployment | 13:13 |
| lyr | what about new installation ? | 13:15 |
| lyr | ah ok, when you say packaged, you mean I'm gonna hit wall after wall installinf sf rpms | 13:15 |
| tristanC | we would have to rebuild the rpm for a new baseos, which shouldn't be too complicated as we have all the dependencies covered | 13:16 |
| tristanC | another strategy we discussed was to publish centos7 based container image, and update sf-config to use podman to install the compoment, that would work on both centos8 and fedora | 13:16 |
| zenkuro | tristanC: second way sounds better, since it can be moved to other systems, and whole sf would be a set of configs | 13:19 |
| zenkuro | I mean if everything in container -> you can focus on protocol and not on the libraries | 13:20 |
| zenkuro | so once everything in container you can slowly upgrade each container/service from centos7 to centos8 | 13:21 |
| tristanC | zenkuro: about the issue you posted yesterday, it seems like the deployment stopped because the telegraf db couldn't be ensure in influx, have been able to solve the error? | 13:21 |
| zenkuro | nope... | 13:22 |
| zenkuro | It looks like if sf is not working for 2-3 week something goes bad... | 13:22 |
| zenkuro | it is a 3-rd time I have this issue | 13:22 |
| zenkuro | I will probably reinstall everything from scratch with x.5 version | 13:23 |
| tristanC | nhicher: do you recall what can cause such issue ^ | 13:23 |
| tristanC | zenkuro: it seems like influxdb is not working, could you check the service log (journalctl -u influxdb) | 13:24 |
| *** jpena|lunch is now known as jpena | 13:29 | |
| zenkuro | tristanC: Nov 12 14:21:01 managesf.ci.open-e.com influxd[1358]: ts=2020-11-12T13:21:01.330604Z lvl=info msg="Retention policy deletion check (end)" log_id=0QPlRxil000 service=retention op_name=retentio | 13:29 |
| zenkuro | tristanC: it starts from yesterday, when I tried to reinstall sf | 13:30 |
| zenkuro | tristanC: and it looks like no error from that moment | 13:30 |
| lyr | tristanC : either way, from user (my) pov being stuck on centos 7 hence sort of python 2 is a pain | 13:33 |
| zenkuro | by the way, it looks like after sfconfig -update all previous logs gets removed | 13:40 |
| tristanC | lyr: what is the issue with python2? sf install the python3.6 from centos base so there should be a python3 setup | 13:41 |
| tristanC | zenkuro: hum that should not happen, which logs are you referring to? | 13:41 |
| tristanC | we have been hesitant to support other base os because of the extra work to maintain and test multiple os, and centos-7 is supported until 2024 | 13:43 |
| zenkuro | tristanC: /var/log/zuul/executor.log /var/log/scheduler.log, or Im confusing it with log rotation. Since new logs a clean | 13:45 |
| lyr | tristanC it's a bit unrelated to zuul. That's my own base setup (dev-sec.os & ssh-hardening, users, ...) which dies on ansible "yum:" tasks. Reason being, Ansible wants either yum + python2 or dnf + python3, otherwise it's missing bindings | 14:00 |
| lyr | So I wish I could give a shot at CentOS 8, "natively" python 3, to see if the problem solve itself | 14:01 |
| tristanC | lyr: it may work for most things that are not build with centos-7 ABI, but i guess package like re2 or haveged will need to be rebuild | 14:02 |
| tristanC | lyr: you can find how the rpms are built in this project, for example the sf-3.5 is defined by https://softwarefactory-project.io/cgit/software-factory/sfinfo/tree/sf-3.5.yaml | 14:06 |
| nhicher | tristanC: no, that's strange we didn't get the issue with telegraf on ci | 14:17 |
| zenkuro | nhicher: like if is working fine and pass tests or it is working fine as you guys use it? | 14:22 |
| nhicher | zenkuro: for each change we propose on sf, we run install and upgrade jobs | 14:23 |
| zenkuro | nhicher: is there possibility that something might go wrong after long period of inactivity? | 14:23 |
| nhicher | zenkuro: I don't think we purge influxdb db, I will check on sf-config codebase | 14:26 |
| zenkuro | if it help anybody: TASK [sf-influxdb : Check if telegraf db exists] --- FAIL | 14:27 |
| zenkuro | and `ls /var/log/telegraf/` returns nothing | 14:28 |
| nhicher | zenkuro: it should not be an issue, we create the db later if it doesn't exist https://softwarefactory-project.io/cgit/software-factory/sf-config/tree/ansible/roles/sf-influxdb/tasks/influxdb_configuration.yml#n27 | 14:37 |
| nhicher | zenkuro: what is the ansible logs after TASK [sf-influxdb : Check if telegraf db exists]? | 14:38 |
| zenkuro | Nov 12 15:38:04 managesf.ci.open-e.com telegraf[1340]: 2020-11-12T14:38:04Z E! [outputs.influxdb] When writing to [https://managesf.ci.open-e.com:8086]: Post "https://managesf.ci.open-e.com:8 | 14:38 |
| zenkuro | Nov 12 15:38:04 managesf.ci.open-e.com telegraf[1340]: 2020-11-12T14:38:04Z E! [agent] Error writing to outputs.influxdb: could not write any address | 14:38 |
| zenkuro | this is a 'journalctl -u telegraf' | 14:39 |
| nhicher | zenkuro: we already got this issue, did you try to restart telegraf? | 14:40 |
| nhicher | zenkuro: when we've got network issue between the 2 hosts | 14:41 |
| zenkuro | nhicher: restarting | 14:41 |
| zenkuro | nhicher: restarted both telegraph and fluxdb, no changes | 14:43 |
| nhicher | zenkuro: it's with centos 7.9? | 14:46 |
| zenkuro | nhicher: Ive updated centos7 to the last current version | 14:46 |
| nhicher | zenkuro: ok, I will test with the last version | 14:46 |
| zenkuro | nhicher: cat /etc/centos-release gives CentOS Linux release 7.8.2003 (Core) | 14:50 |
| zenkuro | hm , I sought that sudo yum update, will give me the latest thing... am I missing something? | 14:51 |
| nhicher | zenkuro: after centos release it takes some time to sync all mirrors | 14:57 |
| zenkuro | nhicher: Ive turned of firewall, no changes | 14:58 |
| nhicher | zenkuro: can you try to connect to influxdb https://softwarefactory-project.io/docs/operator/metrics.html#influxdb-database ? | 15:00 |
| zenkuro | it looks like "influx -ssl -host $influxdb_host -username telegraf -password $telegraf_passwd -database telegraf" is a bit outdated | 15:04 |
| zenkuro | hm | 15:05 |
| zenkuro | influx -ssl | 15:06 |
| zenkuro | Error: unknown shorthand flag: 's' in -ssl | 15:06 |
| zenkuro | nhicher: looks like my local influx behaves differently from manuals | 15:09 |
| zenkuro | nhicher: manually reinstalling influxdb | 15:12 |
| zenkuro | nhicher, tristanC: hurray! \>_</ | 15:30 |
| zenkuro | so the solution is: manualy remove influxdb and telegraph and run sfconfig | 15:30 |
| nhicher | zenkuro: ok, good =) | 15:31 |
| zenkuro | is it possible that influx was corrupted or installed of a wrong version? | 15:32 |
| zenkuro | so will you use containers? | 15:32 |
| tristanC | zenkuro: in sf-3.5 we removed the telegraph part of sf-config, see: https://softwarefactory-project.io/cgit/software-factory/sf-config/tree/releasenotes/notes/telegraf-removal-682d3d701f740c10.yaml | 15:35 |
| tristanC | zenkuro: it seems like containers will be the better solution, but it will takes time as we need to figure out how to create, publish and maintain that new packaging format | 15:36 |
| zenkuro | tristanC: let me know if you guys need help with it. I don't have much spare time but I would be gland to know more about sfactory | 15:38 |
| zenkuro | and help with something | 15:39 |
| erbarr | yo, i'm new to SFP and zuul, following https://www.softwarefactory-project.io/tag/zuul-hands-on-series.html to get familiar and i'm at part 2 and when do the review to run the unittests playbook zuul gets stuck on queued, how can I check what's going on? | 15:55 |
| tristanC | erbarr: queued usually means zuul is waiting for a node, so checking /var/log/nodepool/launcher.log might give some clues | 15:57 |
| erbarr | tristanC, i'm seeing /var/log/nodepool is empty | 15:58 |
| tristanC | erbarr: then that might be the issue, could you check the output of `systemctl status nodepool-launcher` ? | 16:02 |
| erbarr | tristanC, it was dead, so i started it also builder and made an edit to the patch, stuck in queue again and log says nodepool.statsworker starting as the last message | 16:12 |
| tristanC | erbarr: is there a configuration in /etc/nodepool ? | 16:14 |
| erbarr | tristanC, I see nodepool.yaml, logging.conf, secure.conf, logging yaml files for builder and launcher | 16:17 |
| erbarr | some other dirs in there too | 16:18 |
| tristanC | erbarr: in nodepool.yaml, is there a provider? | 16:18 |
| erbarr | tristanC, nope, that's [] | 16:19 |
| tristanC | erbarr: in https://www.softwarefactory-project.io/how-to-setup-a-software-factory-sandbox.html , have you added the `hypervisor` role in the `Install Software Factory` step? | 16:20 |
| tristanC | the hypervisor role configure the hosts and adds a provider to nodepool configuration | 16:21 |
| erbarr | tristanC, ahh, i see, let me add and rerun | 16:23 |
| erbarr | tristanC, thanks, runc was giving me the "providers need to be removed manually" thing so I was looking how to get past that but then saw that runc is deprecated so i changed to k1s and is installing running sfconfig currently, I really will need to change this to have nodepool use openstack, is there something I could look up after i'm done with the hands-on series? | 16:57 |
| tristanC | erbarr: that's correct, you could use k1s instead of runc, but if it's just to get familiar with zuul, then nodepool configuration is not relevant | 17:02 |
| tristanC | erbarr: to use openstack, you first need to setup the cloud provider: https://docs.softwarefactory-project.io/sf-config-3.5/operator/nodepool_operator.html#add-a-cloud-provider | 17:04 |
| tristanC | erbarr: then you would need to configure nodepool like so: https://docs.softwarefactory-project.io/sf-config-3.5/user/nodepool_user.html#nodepool-user | 17:04 |
| erbarr | tristanC, cool thanks! | 17:11 |
| *** apevec has quit IRC | 17:21 | |
| *** apevec has joined #softwarefactory | 17:21 | |
| *** jpena is now known as jpena|off | 17:49 | |
| *** apevec has quit IRC | 19:22 | |
| *** apevec has joined #softwarefactory | 19:37 | |
| *** apevec has quit IRC | 19:42 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!