Chris Ellis @intrbiz@bergamot.social

@giggls refreshing the collation version will make the warning go away. However you might have index corruption, from the underlying changes to the collation.

If you have any type of indexes on text columns, you should reindex those. To rebuild the index aligned with the new collation rules. Then you can refresh the collation version.

@nixCraft also, nothing screams you're employer is awful, than being happy at finding a good candidate under market rate.

A few times, hiring, I found great people who asked for way less than they deserved. I told the company to offer them what they're worth.

The good places did, and we ended up with a great person who felt valued.

The bad places did not, and we ended up with a great person who ended up feeling cheated.

I wish more companies could understand treating people well is good.

@nixCraft getting a job somewhere else. Has universally been my fastest way to a raise, and what I've tended to advise people to do. Employers rarely demonstrate loyalty these days.

I've seen many employers spend more on recruiting a replacement rather than giving a colleague a raise.

The best I once saw, after 3/4 of the team had quit, got a job offer (for comparable pay) then turned around and demanded a 3x raise or they'd leave, leaving the company with no-one in the team, it worked.

@andypiper @codepope @hackwimbledon oh :(

Had many nice conversations with him at PostgreSQL events, will be missed.

@dick_turpin being against things is what the Torry party has done for the last decade.

@amd hmm, but is that not writing 4 copies of the data to the underlying PVs. Which seems a more likely cause of slow performance than it being thin.

@amd I've not experienced significant performance dip due to thinpool on my setup. Wouldn't have thought the slightly extra metadata of thin provisioning would cause a bit impact.

iostat -mx would be worth a look at to see if your saturating the disk IO and where.

When you say raid 1, do you mean raid 10? 4 disks in a mirror seems odd.

Is the slow IO inside a VM or on the host?

@mntmn looks like it's using an RCA socket with a switched terminal and wiring the switched part to the other channel. I'm guessing, but looks like it's trying to operate in mono if only one RCA plug is plugged in, but in stereo if both are plugged in.