Hello,
I am considering Wednesday 27th (next week) at 14:00 UTC to do this change. If no one says otherwise, I will go ahead and schedule it.
Thanks Manuel.
On Wed, Oct 20, 2021 at 8:17 AM Manuel Arostegui marostegui@wikimedia.org wrote:
On Wed, Oct 20, 2021 at 1:23 AM Kunal Mehta legoktm@wikimedia.org wrote:
Hi,
On 10/19/21 6:00 AM, Manuel Arostegui wrote:
My idea is to decrease the current TTL from m5-master (from 5 minutes
to
1 minute, so we can revert faster if needed). Once that is in place, I will simply change the CNAME to point to dbproxy1017.eqiad.wmnet. Everything should be transparent (if the
grants
are correct), and connections should smoothly change to the proxy in
the
next few days unless you want to restart the applications right away.
Sounds good. I think it's probably just easier for us to restart mailman and mailman-web soon after to find any issues right away than later on.
-- Kunal
Sounds good! I will email this thread and update the task once I've decided a day/time for this change for next week.
Hello,
Calendar invite sent
On Thu, Oct 21, 2021 at 8:56 AM Manuel Arostegui marostegui@wikimedia.org wrote:
Hello,
I am considering Wednesday 27th (next week) at 14:00 UTC to do this change. If no one says otherwise, I will go ahead and schedule it.
Thanks Manuel.
On Wed, Oct 20, 2021 at 8:17 AM Manuel Arostegui marostegui@wikimedia.org wrote:
On Wed, Oct 20, 2021 at 1:23 AM Kunal Mehta legoktm@wikimedia.org wrote:
Hi,
On 10/19/21 6:00 AM, Manuel Arostegui wrote:
My idea is to decrease the current TTL from m5-master (from 5 minutes
to
1 minute, so we can revert faster if needed). Once that is in place, I will simply change the CNAME to point to dbproxy1017.eqiad.wmnet. Everything should be transparent (if the
grants
are correct), and connections should smoothly change to the proxy in
the
next few days unless you want to restart the applications right away.
Sounds good. I think it's probably just easier for us to restart mailman and mailman-web soon after to find any issues right away than later on.
-- Kunal
Sounds good! I will email this thread and update the task once I've decided a day/time for this change for next week.
This was a failure, I had to revert since mailman, toolhub and striker were failing (I fixed striker on the fly - grants missing) The others seems to be a firewall/connectivity issue - I will investigate and follow up on the task.
Thank you Andrew, Bryan and Kunal for taking the time to help me with testing (and sorry for breaking things!)
On Fri, Oct 22, 2021 at 7:02 AM Manuel Arostegui marostegui@wikimedia.org wrote:
Hello,
Calendar invite sent
On Thu, Oct 21, 2021 at 8:56 AM Manuel Arostegui marostegui@wikimedia.org wrote:
Hello,
I am considering Wednesday 27th (next week) at 14:00 UTC to do this change. If no one says otherwise, I will go ahead and schedule it.
Thanks Manuel.
On Wed, Oct 20, 2021 at 8:17 AM Manuel Arostegui < marostegui@wikimedia.org> wrote:
On Wed, Oct 20, 2021 at 1:23 AM Kunal Mehta legoktm@wikimedia.org wrote:
Hi,
On 10/19/21 6:00 AM, Manuel Arostegui wrote:
My idea is to decrease the current TTL from m5-master (from 5 minutes
to
1 minute, so we can revert faster if needed). Once that is in place, I will simply change the CNAME to point to dbproxy1017.eqiad.wmnet. Everything should be transparent (if the
grants
are correct), and connections should smoothly change to the proxy in
the
next few days unless you want to restart the applications right away.
Sounds good. I think it's probably just easier for us to restart mailman and mailman-web soon after to find any issues right away than later on.
-- Kunal
Sounds good! I will email this thread and update the task once I've decided a day/time for this change for next week.
cloud-admin@lists.wikimedia.org