showing posts flaired Maintenance — clear

0

1 hour outage coming up sometime in the next several days

preview

Sometime between Friday evening and Sunday late evening (EDT), there will be a 1-hour outage on feddit.online so I can upgrade the server from Ubuntu 24.04 LTS to 26.04 LTS.

Ubuntu releases a new LTS version every 2 years, and so for one hour every 2 years, I need to upgrade Ubuntu.

Sometime between Friday evening and Sunday late evening (EDT), there will be a 1-hour outage on feddit.online so I can upgrade the server from Ubuntu 24.04 LTS to 26.04 LTS.

Ubuntu releases a new LTS version every 2 years, and so for one hour every 2 years, I need to upgrade Ubuntu.

0

Digital Ocean warns of 3 days of disruptions

preview

I received the following maintenance notification from Digital Ocean today (27-July-2026) warning of upcoming disruptions in service. Note that while they say things should recover, from experience, this has not always happened. An outage could last longer if things do not automatically recover.

During the above window, our Networking team will be making changes to the core networking infrastructure to improve performance and scalability for a subset of Droplets in the NYC1 region.

This maintenance will be carried out in three phases as outlined below:

August 3, 09:00 to 11:00 UTC,
August 4, 09:00 to 11:00 UTC,
August 5, 09:00 to 11:00 UTC

Expected impact:

This operation will cause a very short traffic interruption. The traffic should recover automatically afterwards. In case any prolonged impact is noticed, the change will be reverted immediately. We will endeavor to keep this to a minimum for the duration of the change.

I received the following maintenance notification from Digital Ocean today (27-July-2026) warning of upcoming disruptions in service. Note that while they say things should recover, from experience, this has not always happened. An outage could last longer if things do not automatically recover.

During the above window, our Networking team will be making changes to the core networking infrastructure to improve performance and scalability for a subset of Droplets in the NYC1 region.

This maintenance will be carried out in three phases as outlined below:

August 3, 09:00 to 11:00 UTC,
August 4, 09:00 to 11:00 UTC,
August 5, 09:00 to 11:00 UTC

Expected impact:

This operation will cause a very short traffic interruption. The traffic should recover automatically afterwards. In case any prolonged impact is noticed, the change will be reverted immediately. We will endeavor to keep this to a minimum for the duration of the change.

0

DB scaling up is complete

preview

2025-Mar-14: I scaled up the DB cluster from regular SSD drives to nVME drives, and moved from regular Intel CPUs to Premium AMD CPUs. nVME is significantly faster in all metrics than regular SSD drives. The improvement should be significant.

The load values are much lower than they were before, so, so far, so good.

Please let me know if you notice any issues.

2025-Mar-14: I scaled up the DB cluster from regular SSD drives to nVME drives, and moved from regular Intel CPUs to Premium AMD CPUs. nVME is significantly faster in all metrics than regular SSD drives. The improvement should be significant.

The load values are much lower than they were before, so, so far, so good.

Please let me know if you notice any issues.

0

Scaling up the DB cluster this evening 14-Mar-2025

preview

I'm seeing signs that the DB Cluster is struggling at times and during peak activity it may become a bottleneck. The application server is cruising.

This evening, I'm going to scale up the DB cluster from regular SSD drives to nVME drives, and move from regular Intel CPUs to Premium AMD CPUs. nVME is significantly faster in all metrics than regular SSD drives. The improvement should be significant.

There shouldn't be any downtime unless something goes wrong.

I'm seeing signs that the DB Cluster is struggling at times and during peak activity it may become a bottleneck. The application server is cruising.

This evening, I'm going to scale up the DB cluster from regular SSD drives to nVME drives, and move from regular Intel CPUs to Premium AMD CPUs. nVME is significantly faster in all metrics than regular SSD drives. The improvement should be significant.

There shouldn't be any downtime unless something goes wrong.