TZ: UTC +1/+2
This is an interesting finding, maybe we need to pass this info along to Marko and see if they can guess why (and if it is expected) it happens.
Also interesting to see that the script is faster than this last run of myloader - good job! :)
metawiki, mediawikiwiki, labswiki,officewiki aren't having echo_unread_wikis table anymore.
So this is only needed in x1 (wikishared)
I am going to start working on this in January. It will require x1 to go to SBR though for a few days.
@KHurd-WMF this is not yet done - I was just verifying it is now fine and also added you to the Phabricator group wmf-nda
This can now proceed and can be picked up by the clinic duty person this week!
So, check_user looks good and KHurd1 is associated to khurd WMF email account now.
@KHurd-WMF Thanks for the explanation. It is probably easier if you keep KHurd1 then as it is associated to your wmf email account already. Could you edit the task to reflect that's the user that needs to go into the wmf group?
@sbassett I am not sure KHurd is the right user name, from what I can see there are two users with KHurd, there is KHurd and KHurd1, both created in 2022. One was created in Nov 2022 and another one yesterday (6th Dec).
Only KHurd1 has a wikimedia email associated to.
@Krinkle can this be closed?
This host is now being back to serve traffic.
@Arnoldokoth let's discuss here what you need.
Do you need a new user with just SELECT grant?
All done from the DBA side.
email@example.com[(none)]> select user,host from mysql.user where host like '10.64.16.8'; +----------------+------------+ | User | Host | +----------------+------------+ | phabricatorphd | 10.64.16.8 | | phadmin | 10.64.16.8 | | phmanifest | 10.64.16.8 | | phstats | 10.64.16.8 | | phuser | 10.64.16.8 | +----------------+------------+
I will merge that change and then proceed and remove grants live
That's ok Daniel, I will take care of it on this task.
Mon, Dec 5
We can give it only SELECT grant
Is there a way to monitor this so we can get an alert if it happens again?
it is now pulled with the normal weight. I will leave it during working hours to see how it goes and later depool it again.
Ah cool, so the times are pretty similar, and there's not a big difference (for good or bad).
The RAID controller itself looks good.
The default options we care about are all present:
- Learning cycles disabled
- RAID10 was done by default
- Strip size 256k
- No read ahead
After replicating fine during the last few days - I have pooled this host with just 1% weight to see how the controller does in terms of performance. I will slowly increase its weight during the day.
Host being repooled automatically.
Fri, Dec 2
Excellent, closing this then!
Added to nda group and created the kerberos principal. @dasm you should've received an email with further instructions. Also please allow 30-60 minutes for puppet to run everywhere.
Please reopen if you have troubles accessing.
Thanks for the update! Looking forward to seeing that next comparison
@jcrespo do you want/have a tracking task to productionize these hosts?
Thu, Dec 1
I have merged your patch. Also, you should've gotten an email about your kerberos principal.
Please also allow 30-60 minutes for puppet to spread across the fleet.
Merged and applied the grants. Please test it and reopen if it is not working!
@matthiasmullie do you want to also add yourself to analytics-privatedata-users in the gerrit patch? Once done I can merge and add the kerberos principal too.
ssh key verified and replaced. Please allow 30-60 minutes for it to totally spread across the fleet.
Thanks Greg!. I will wait for the correct template and then proceed.
db1206 got installed correctly and worked out of the box.
However, trying to play around with megacli has revealed that the host gets frozen when running some of the commands to get controller's info.
@andrea.denisse - can you merge and submit your patch so we can create the kerberos principal and close this task?
I am going to close this, please reopen if adding you to the WMF group wasn't enough.
Applied to db2173 which was down as part of T322988
Thank you Papaul!
Applied to db2173 which was down as part of T322988
Wed, Nov 30
Thanks @Damilare - which access do you need? https://wikitech.wikimedia.org/wiki/Analytics/Data_access#What_access_should_I_request?
You just need Turnilo without accessing PII? It is not clear what you'd need :)
I will hold until the group needed is sorted :)
It would be interesting to see the difference in length, in a sX section between that script and myloader. Just to have some rough idea...
Confirmed L3 is signed.
@MarkTraceur we need your approval for this.
@VirginiaPoundstone you were not in the WMF LDAP group. I just added you. Can you retry?
This is done - please give it 30-60 minutes for the change to spread everywhere.
Hello @Damilare can you please follow the proper template at: https://phabricator.wikimedia.org/maniphest/task/edit/form/8/
Hello @AnnWF can you please follow the proper template at: https://phabricator.wikimedia.org/maniphest/task/edit/form/8/
Tue, Nov 29
Is this something @Papaul can finish?
I think this might be more specific for these two teams.
Thank you Jaime!