Service Restored from Overnight Power Outage

Yesterday, the Internet Archive experienced a power outage that took some of our services offline and degraded others. The power outage at one of our datacenters was caused by damage to the power feed, necessitating PG&E to repair it.
We are still bringing up the upload feature, but otherwise we are live and kicking. Please enjoy the old stuff.
PG&E crews arrived and worked until 4 a.m., alongside our infrastructure staff, to restore power. By 4:15 a.m., our fans, network, and servers were coming back online in sequence.
A big thank you to the PG&E crew and our staff for working through the night to restore services and get the Internet Archive back online.
Always the last mile.
The problems now are just a very slow and error prone on most files attempted. One file opens while another one won’t. Maybe an hour later that one will open and another won’t. So while it tries to open and after so many seconds just gives off various errors. Usually when clicking on a file its like a second or less and opens. Now its like 10 seconds if it opens or it just fails after so much time. Basically half or more of my actual Uploads fail along with others on IA. It has been like this for days. Reading the DL (failure) charts one can see peaks and valleys hours in. Just not stable from hour to hour. I know some are working 7/24 on this and appreciate it. Those broken power line polls look aged, maybe decades? Don’t think I’ve seen anything like that around here. Hopefully it gets repaired. You now have a Trillion + web pages. Keep up the great work that you do which is phenomenal!
It’s seem that the sites is not 100 percent up due to glitches on the sites that be cuase by that power outage this past weekend hopefully you guys here getting it fix a bit so keep us posted on this please

Facts Only

* The Internet Archive experienced a power outage at one data center.
* The cause was damage to the power feed.
* PG&E was necessitated to repair the power feed.
* PG&E crews and infrastructure staff worked to restore power until 4 a.m.
* Fans, network, and servers came back online in sequence by 4:15 a.m.
* File attempts experienced slow loading times or errors following the outage.
* A significant number of uploads failed alongside other issues.
* Failure charts showed unstable performance over time with peaks and valleys.

Executive Summary

The Internet Archive experienced a power outage at one of its data centers, which resulted in degraded service availability. The outage was caused by damage to the power feed, requiring PG&E to perform repairs. Restoration involved PG&E crews and infrastructure staff working until 4 a.m., with systems bringing back online sequentially. Following the restoration, users reported significant issues with file access, noting that many uploads failed or were slow and error-prone. Reports indicated instability in file opening times, with some files failing repeatedly while others opened quickly. The situation is ongoing, with reports suggesting incomplete service recovery due to lingering glitches on some sites.

Full Take

The narrative balances an initial successful physical restoration effort with a subsequent, persistent functional degradation affecting data access and service stability. The immediate action involves a clear delineation of responsibility for the outage (PG&E) and recovery (staff), which establishes a concrete operational event. However, the persistence of file access errors introduces a secondary layer of instability that supersedes the physical repair narrative. The shift from a singular, time-bound incident to a prolonged state of inconsistent performance highlights a critical gap: the perceived success of restoring power did not equate to restored system integrity or functional reliability for end-users. This suggests that systemic resilience is not solely about hardware recovery but also involves ensuring data consistency and operational stability post-event. The undercurrent hints at questions regarding the dependency of digital infrastructure on external, vulnerable physical systems, and whether recovery protocols adequately account for subsequent software/service glitches introduced during or after the disruption.

Sentinel — Human

Confidence

The text exhibits strong human markers through its informal tone, subjective experience, and fluctuating register, suggesting it is a direct communication rather than formal journalism.

Signals Detected

Sentence length variance is erratic; tone shifts from formal announcement to informal reflection.

Fluency is interrupted by highly subjective, immediate observations and personal anecdotes regarding file access failures and infrastructure age.

No discernible pattern of argumentative structure; the text functions more as an internal status update/complaint than a formal news report.

The inclusion of highly specific, unverified personal observations ('Those broken power line polls look aged, maybe decades?') suggests a direct, immediate experience rather than polished reporting.

Human Indicators

Use of highly informal and subjective language (e.g., 'live and kicking', 'old stuff', 'broken power line polls').

Direct expressions of frustration regarding technical instability ('one can see peaks and valleys hours in', 'so many seconds just gives off various errors').

A combination of formal acknowledgment (thanking crews) layered with highly personal, conversational commentary.