BradC

> If finalized, this action will devastate the EPA’s ability to carry out its primary authority to limit climate pollution under the federal Clean Air Act.

holy cow

> Reuters reported last week that the EPA plans to repeal all greenhouse gas emission standards for light-duty, medium-duty, and heavy-duty vehicles and engines in the coming days after it removes the scientific finding that justified those rules, according to a summary.

Trump's EPA to repeal finding that climate pollution endangers human health

> The justice department has charged at least 26 people with “assaulting” and “impeding” federal officers and other crimes during the protests over immigration raids. Prosecutors, however, have since been forced to dismiss at least eight of those felonies, many of them which relied on officers’ inaccurate reports, court records show.

> US immigration officers made false and misleading statements in their reports about several Los Angeles protesters they arrested during the massive demonstrations that rocked the city in June, according to federal law enforcement files obtained by the Guardian.

Dropped cases against LA protesters reveal false claims from federal agents

> The three officials who spoke to The Times on condition of anonymity said prosecutors have struggled to get several protest-related cases past grand juries, which need only to find probable cause that a crime has been committed in order to move forward. That is a much lower bar than the “beyond a reasonable doubt” standard required for a criminal conviction.

LA Grand Juries Are Refusing to Indict ICE Protestors

Wisconsin Supreme Court’s liberal majority strikes down 176-year-old abortion ban

Trump’s justice department issues directive to strip naturalized Americans of citizenship for civil offenses

(Insert Trump corruption meme here)

> As for the network, Trump Mobile’s online Terms of Use say that it’s “powered by Liberty Mobile,” a freedom-themed mobile virtual network operator (MVNO) that uses the T-Mobile network.

> There are numerous errors on the page, from a processor section that doesn’t list a processor, RAM that’s described as storage, and the boast of a “5000mAh long life camera,” when it presumably means the battery.

> Details on the phone itself are sketchy. The product page includes what appears to be a Photoshopped image of a brushed gold phone with a “T1” logo and US flag on the back. The triple camera layout appears similar to recent iPhone Pro models at first glance, but the lenses are strangely spaced, and there doesn’t appear to be any camera flash.

Trump Mobile launches $47 service and a gold phone

> The shocking scene of a U.S. senator being aggressively removed from a Cabinet secretary’s news conference prompted immediate outrage from his Democratic colleagues. Images and video of the scuffle ricocheted through the halls of Congress, where stunned lawmakers demanded an immediate investigation and characterized the episode as another in a line of mounting threats to democracy by President Donald Trump’s administration.

Sen. Padilla is forcefully removed from Noem’s news conference on immigration raids and handcuffed

> Climate.gov, the gateway website for the National Oceanic and Atmospheric Administration (Noaa)’s Climate Program Office, will imminently no longer publish new content, according to multiple former staff responsible for the site’s content whose contracts were recently terminated.

Major US climate website likely to be shut down after almost all staff fired

The MAHA Report Cites Studies That Don’t Exist

I'm 84% certain that the only difference in size on disk between 1 log backup hourly and 60 log backups in the same hour is a small header in each file

My understanding is that (within a margin of error) you're going to end up with the same GB of tran logs whether you take them hourly, every 15 min or every minute

@J.Mini I don't think that's true; don't tran log backups only fully capture committed transactions?

hence the move to diffs

(but would still require the full + huge diff to restore)

hmm, good point. I mean if the entire database was changed, its still going to only max out at the size of a full

you'd rather do full + 5 days of log than full + diff + 1 day of log?

(not to mention the complication of maintenance like reindexing)

but we do have a couple of dbs that see regular data dumps and imports and processing, which might be exceptions to the low daily change we see for most apps

yeah, log backups are over and above all this discussion, which is mostly about keeping nightly maintenance windows minimized and reducing back-end storage needs

so I'll be taking a new full because of retention policy far sooner than I'll be taking a full because the diff gets too big

its highly likely this is entirely moot, if the vast majority of my dbs have, like a max 2% daily change rate

eh. I'll just track the actuals as well as the actual backup sizes and see what proves out

so modified/allocated seems more predictive of diff size than modified/total (assuming backup compression is on)

I guess there might be some weird outliers where the modified extents ARE extents that become unallocated by deletes or something, but I can't imagine those are common

(by that I mean "logic performed in advance", not "highly advanced logic")

Well, Veeam doesn't do this, that's why I have to do some advanced logic before calling out to the Veaam SQL plug-in executable

@Charlieface well, I'm writing the logic and then explicitly calling a full or a differential, so its on me

(since unallocated extents will be compressed to near-zero)

and if it works well enough, it could even mean that low-activity dbs could do a full every 2 weeks and a tiny diff nightly

I'd rather put in some logic that checks in advance

I just don't want to schedule weekly fulls and nightly diffs then discover later that one super busy database has diffs that grow super huge by day 3

lolol

Is there a "right" percent of size that should lean me toward escalating a diff to a full? 50%? 75%?

But in fact that's probably true for a diff that's 90% or 80% of the size of the full; I should be doing a full instead