[openstack-community] [services] Today's skipped meeting, moving the conversation forward
Hello folks, today we skipped the meeting, I messed up with the times. Anyway, some of the items are just status updates on ongoing tasks and I think we can keep the conversation open on the mailing list. https://wiki.openstack.org/wiki/Community/MeetingAgenda#Community_Software_S... * dizquierdo and dpose to keep working on the https://etherpad.openstack.org/p/activity-infratization-spec file and the architecture diagram * dpose and dizquierdo to keep working on the puppet recipes at https://github.com/MetricsGrimoire/puppet-metricsgrimoire * dizquierdo will check the vagrant machine to have a look at the internal machinery of grimoire toolset * add start-date end-date to event xml, so tipit can consume it https://storyboard.openstack.org/#!/story/2000193 * reed to check back with fungi in a couple of days, re https://storyboard.openstack.org/#!/story/2000158, Askbot infratization (DONE, wasn't needed) The most pressing issue is https://review.openstack.org/#/c/160693, which seems to have made a lot of progress already. What steps do we need to follow to test if the new Askbot machine is working correctly? Thanks, stef PS: I renamed the meeting into Community Software Services so it doesn't clash with the other project called OpenStack Infrastructure]
On 2015-03-16 10:08:21 -0700 (-0700), Stefano Maffulli wrote: [...]
reed to check back with fungi in a couple of days, re https://storyboard.openstack.org/#!/story/2000158, Askbot infratization (DONE, wasn't needed)
Thanks--I missed that story, so I've updated assignments and completion status for the tasks there.
The most pressing issue is https://review.openstack.org/#/c/160693, which seems to have made a lot of progress already. What steps do we need to follow to test if the new Askbot machine is working correctly?
Good question. Presumably just edit your local /etc/hosts file to make ask.openstack.org resolve as 104.239.140.123 (or 2001:4800:7819:104:be76:4eff:fe05:b1be if you want to test drive IPv6 for it) and see what's working. The snapshot I imported is current as of three days ago, but we'll wipe it and do one more import during a scheduled outage once everyone's satisfied it's working as intended.
I renamed the meeting into Community Software Services so it doesn't clash with the other project called OpenStack Infrastructure
Thanks! That should work great. -- Jeremy Stanley
On Mon, 2015-03-16 at 17:17 +0000, Jeremy Stanley wrote:
Good question. Presumably just edit your local /etc/hosts file to make ask.openstack.org resolve as 104.239.140.123 (or 2001:4800:7819:104:be76:4eff:fe05:b1be if you want to test drive IPv6 for it) and see what's working.
Thanks Jeremy. This means we're very close to completing this task. I have modified my etc/hosts and the site seems to be working just fine: I can successfully search, edit questions, see the list of people ... I don't see any issue. I didn't even get any complaints from the SSL certificates :) Smooth as oil, as Italians say. Marton, Evgeny: are there any specific things we should be looking for specifically to make sure that the new server works?
The snapshot I imported is current as of three days ago, but we'll wipe it and do one more import during a scheduled outage once everyone's satisfied it's working as intended.
How long should we plan for the outage to be? 2-3 hours or less? Thanks, stef
On 2015-03-16 10:57:33 -0700 (-0700), Stefano Maffulli wrote: [...]
I didn't even get any complaints from the SSL certificates :) Smooth as oil, as Italians say. [...]
You might want to do something to double-check that you're actually interacting with the new server rather than the old one (for example make some edits, then switch back to the old server and confirm you don't see them there). But correct, you shouldn't see complaints about the SSL certificate--the same one is installed on both old and new servers and your browser validates it based on the hostname you've entered, not based on any particular IP address.
How long should we plan for the outage to be? 2-3 hours or less?
Having run through it once myself already, it should be doable in under 30 minutes (most of that time is spent waiting for the database dump, file transfer, import and search reindexing). -- Jeremy Stanley
On Mon, 2015-03-16 at 10:57 -0700, Stefano Maffulli wrote:
Marton, Evgeny: are there any specific things we should be looking for specifically to make sure that the new server works?
Here are a few tests we should do. Please change your /etc/hosts file adding this line: 104.239.140.123 ask.openstack.org and if your host is IPv6 capable: 2001:4800:7819:104:be76:4eff:fe05:b1be ask.openstack.org Test cases: - Login/logout with Launchpad - Login/logout with Google - Login/logout with Yahoo - Signup new account with Launchpad|Google|Yahoo - Ask a question - Vote for questions and answers - Add a comment, delete a comment - Edit a question PASS (edited so that it wouldn't send alerts) - Change settings (admin only) - Use Search (results are shown, in English) These all PASS for me, I received the email notifications and all. - Search in Chinese FAIL https://ask.openstack.org/zh/questions/scope:all/sort:activity-desc/page:1/q... https://ask.openstack.org/zh/questions/scope:all/sort:activity-desc/page:1/q... The Chinese UI of the site seems to be slightly borked, the search bar goes off to the right (but this is also borked on the old site, not sure when that broke). Marton, Evgeny, Jeremy: can you help diagnose why search in Chinese fails? thanks, stef
On 2015-03-18 16:16:34 -0700 (-0700), Stefano Maffulli wrote: [...]
Marton, Evgeny, Jeremy: can you help diagnose why search in Chinese fails?
After the import I ran `python manage.py askbot_rebuild_index -l zh` (as well as "en") and no errors were returned. If there are diagnostic steps I should perform locally, I'm happy to report the results. -- Jeremy Stanley
I can test this on the vagrant instance, so if I can't reproduce the Chinese issue there, then we can go further with the investigation at infra side. Marton 2015-03-18 17:08 GMT-07:00 Jeremy Stanley <fungi@yuggoth.org>:
On 2015-03-18 16:16:34 -0700 (-0700), Stefano Maffulli wrote: [...]
Marton, Evgeny, Jeremy: can you help diagnose why search in Chinese fails?
After the import I ran `python manage.py askbot_rebuild_index -l zh` (as well as "en") and no errors were returned. If there are diagnostic steps I should perform locally, I'm happy to report the results. -- Jeremy Stanley
_______________________________________________ Community mailing list Community@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/community
On 2015-03-19 11:39:06 -0700 (-0700), Marton Kiss wrote:
I can test this on the vagrant instance, so if I can't reproduce the Chinese issue there, then we can go further with the investigation at infra side.
Also in another message Evgeny suggested I check for the cron jobs and backup dumps on the new server, and I don't see that those have been implemented in the Puppet module yet. Can you confirm? -- Jeremy Stanley
On Thu, 2015-03-19 at 18:42 +0000, Jeremy Stanley wrote:
Also in another message Evgeny suggested I check for the cron jobs and backup dumps on the new server, and I don't see that those have been implemented in the Puppet module yet. Can you confirm?
Since the patch to fix solr and its index merged, I have ran another check. Search in Chinese works only for roman characters, not with Chinese character set. Search is broken on the main site too, and the Chinese UI is also broken similarly. I'd say let's move to the new site and fix it on the new server. Thoughts? /stef
You're right I was testing with roman characters only. So it was a bug in the testing methodology. We need to check how it is passing the special non-roman characters to solr. Chinese indexing is a bit tricky, and this is the reason that Solr have a special analyser for that. Strange that Chinese users never complained. M. 2015-03-27 16:53 GMT+01:00 Stefano Maffulli <stefano@openstack.org>:
On Thu, 2015-03-19 at 18:42 +0000, Jeremy Stanley wrote:
Also in another message Evgeny suggested I check for the cron jobs and backup dumps on the new server, and I don't see that those have been implemented in the Puppet module yet. Can you confirm?
Since the patch to fix solr and its index merged, I have ran another check.
Search in Chinese works only for roman characters, not with Chinese character set. Search is broken on the main site too, and the Chinese UI is also broken similarly.
I'd say let's move to the new site and fix it on the new server.
Thoughts?
/stef
_______________________________________________ Community mailing list Community@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/community
On 16/03/15 18:08, Stefano Maffulli wrote:
Hello folks,
today we skipped the meeting, I messed up with the times. Anyway, some of the items are just status updates on ongoing tasks and I think we can keep the conversation open on the mailing list.
https://wiki.openstack.org/wiki/Community/MeetingAgenda#Community_Software_S...
* dizquierdo and dpose to keep working on the https://etherpad.openstack.org/p/activity-infratization-spec file and the architecture diagram * dpose and dizquierdo to keep working on the puppet recipes at https://github.com/MetricsGrimoire/puppet-metricsgrimoire Hi there,
We updated the architecture file at [1] and added a new Puppet recipe for Mailing List Stats at [2].
* dizquierdo will check the vagrant machine to have a look at the internal machinery of grimoire toolset
With this respect this is available at [3] and this may help to understand how the Grimoire toolset works. As further work we'd like to check how to start dealing with the potential issues of updating databases schemas, issues related to inconsistencies in the database and how to fix them, and public access to those. Thanks for the email! Regards, Daniel. [1] https://raw.githubusercontent.com/MetricsGrimoire/puppet-metricsgrimoire/mas... [2] https://github.com/MetricsGrimoire/puppet-metricsgrimoire/tree/master/Mailin... [3] https://github.com/VizGrimoire/VizGrimoireUtils/tree/master/vagrant
* add start-date end-date to event xml, so tipit can consume it https://storyboard.openstack.org/#!/story/2000193 * reed to check back with fungi in a couple of days, re https://storyboard.openstack.org/#!/story/2000158, Askbot infratization (DONE, wasn't needed)
The most pressing issue is https://review.openstack.org/#/c/160693, which seems to have made a lot of progress already. What steps do we need to follow to test if the new Askbot machine is working correctly?
Thanks, stef
PS: I renamed the meeting into Community Software Services so it doesn't clash with the other project called OpenStack Infrastructure]
_______________________________________________ Community mailing list Community@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/community
-- Daniel Izquierdo Cortazar, PhD Chief Data Officer --------- "Software Analytics for your peace of mind" www.bitergia.com @bitergia
On Tue, 2015-03-17 at 00:13 +0100, Daniel Izquierdo wrote:
We updated the architecture file at [1]
Great, thank you.
and added a new Puppet recipe for Mailing List Stats at [2].
Cool, let's put this for review next week.
* dizquierdo will check the vagrant machine to have a look at the internal machinery of grimoire toolset
With this respect this is available at [3] and this may help to understand how the Grimoire toolset works.
As further work we'd like to check how to start dealing with the potential issues of updating databases schemas, issues related to inconsistencies in the database and how to fix them, and public access to those.
Let's add this as a topic for next week's meeting.
[3] https://github.com/VizGrimoire/VizGrimoireUtils/tree/master/vagrant
/stef
participants (4)
-
Daniel Izquierdo
-
Jeremy Stanley
-
Marton Kiss
-
Stefano Maffulli